Multimodal knowledge graphs (MKGs) organize multimodal facts in the form of entities and relations, and have been successfully applied to several downstream tasks. Since most MKGs are incomplete, the MKG completion (MKGC) task has been proposed to address this problem, which aims to complete missing entities in MKGs. Previous most works obtain reasoning ability by capturing the correlation between target triplets and related images, but they ignore contextual semantic information and the reasoning process is not easily explainable. To address these issues, we propose a novel text-enhanced transformer fusion network called TE-TFN, which converts the context path between head and tail entities into natural language text and fuses multimodal features from both coarse and fine granularities through a multi-granularity fuser. It not only effectively enhances text semantic information, but also improves the interpretability of the model by introducing paths. Experimental results on benchmark datasets demonstrate the effectiveness of our model.

The proliferation in embedded and communication technologies made the concept of the Internet of Medical Things (IoMT) a reality. Individuals' physical and physiological status can be constantly monitored, and numerous data can be collected through wearable and mobile devices. However, the silo of individual data brings limitations to existing machine learning approaches to correctly identify a user's health status. Distributed machine learning paradigms, such as federated learning, offer a potential solution for privacy-preserving knowledge sharing without sending raw personal data. However, federated learning is vulnerable to harmful participants that can degrade the overall model quality by sharing low-quality data. Therefore, it is critical to select suitable participants to ensure the accuracy and efficiency of federated learning. In this article, a unique clustering-based approach is proposed to use social context data for participant selection. Different edge participant groups will be established, and group-specific federated learning will be performed. The models of various edge groups will be further aggregated to strengthen the robustness of the global model. The experimental results demonstrated that through participant selection, clustering-based hierarchical federated learning can achieve better results with less participants in two different IoMT applications for ECG and human motion monitoring. This shows the efficacy of the proposed method in improving federated learning performance and efficiency in various IoMT applications.

Precise and rapid categorization of images in the B-scan ultrasound modality is vital for diagnosing ocular diseases. Nevertheless, distinguishing various diseases in ultrasound still challenges experienced ophthalmologists. Thus a novel contrastive disentangled network (CDNet) is developed in this work, aiming to tackle the fine-grained image categorization (FGIC) challenges of ocular abnormalities in ultrasound images, including intraocular tumor (IOT), retinal detachment (RD), posterior scleral staphyloma (PSS), and vitreous hemorrhage (VH). Three essential components of CDNet are the weakly-supervised lesion localization module (WSLL), contrastive multi-zoom (CMZ) strategy, and hyperspherical contrastive disentangled loss (HCD-Loss), respectively. These components facilitate feature disentanglement for fine-grained recognition in both the input and output aspects. The proposed CDNet is validated on our ZJU Ocular Ultrasound Dataset (ZJUOUSD), consisting of 5213 samples. Furthermore, the generalization ability of CDNet is validated on two public and widely-used chest X-ray FGIC benchmarks. Quantitative and qualitative results demonstrate the efficacy of our proposed CDNet, which achieves state-of-the-art performance in the FGIC task.

Safety production surveillance is of great significance to industrial operation management. While augmented intelligence of things is demonstrating tremendous potential in industrial applications, the analyzed information offers lots of benefits to the higher level planning in the enterprise management systems, to further improve the operational efficiency. In this article, a video surveillance system with augmented intelligence of things is considered as a promising solution to enhance the operational efficiency of enterprises. However, the challenge is to process the surveillance video streams as soon as possible without ignoring any emergencies. This issue can be formulated as a two-stage scheduling problem, which is an NP-hard problem that can be integrated with higher level enterprise systems for operational efficiency improvement. An improved Deep Q-Network (DQN) model with a newly designed prioritized replay scheme, named Bi-Dueling DQN with Prioritized Replay, is proposed to solve this two-stage scheduling problem in a smart enterprise management system. A dense reward function based on a concrete state representation is designed to tackle the sparse reward challenge and to speed up the convergence in actual large-scale task scheduling process. A prioritized replay scheme is then developed to improve the sampling efficiency, so as to effectively reduce the training time in deep reinforcement learning for the optimal two-stage scheduling. The experiment results demonstrated that the proposed approach is able to provide an efficient scheduling policy to resolve the two-stage scheduling problem, while at the same time offering insight information to improve the performance of higher level smart enterprise management systems.

Modern manufacturing process is commonly composed of multiple automated devices working together efficiently. Cloud-based manufacturing aims to achieve better efficiency by allowing the collaborative manufacturing across a group of automated robots. Cooperations between multiple robots can accomplish more complicated tasks that is beyond the capability of any individual one. However, it is of critical importance to control robots with different capabilities to work in harmony while ensuring safety and reliability during this process. In this paper, a double constrained containment mechanism is proposed to dispatch heterogeneous robots in a distributed containment control framework for smart manufacturing. Following a three-layer control framework, a cloud decision-making center is designed to realize the cloud-based collaborative manufacturing, which is more cost-effective than other commonly used containment mechanisms that only rely on information exchange among leader-robots. A projection-based containment control scheme is developed, which not only consider nonlinearities induced by position constraints and velocity constraints, but also can tackle dynamically changing communication topologies with uncertain communication delay, to efficiently navigate all follower-robots into a safe working zone formed by leaders. A theoretical stability analysis is conducted to prove the proposed mechanism can ensure all followers enter the target area while their positions and velocities remaining in the corresponding constraint sets. Experiment evaluation results under three application scenarios demonstrate the advantage of our method that can offer a more practical solution to other existing multi-robot containment control for cloud-based smart manufacturing, in considering both position and velocity constraints combined with switching topologies and communication delays.

Swarms of mobile robots are being widely applied for complex tasks in various practical scenarios toward modern smart industry. Federated learning (FL) has been developed as a promising privacy-preserving paradigm to tackle distributed machine learning tasks for mobile robotic systems in 5G and beyond networks. However, unstable wireless network conditions of the complex and harsh working environment may lead to poor communication quality and bring big challenges to traditional centralized global training in FL models. In this article, a Peer-to-Peer (P2P) based Privacy-Perceiving Asynchronous Federated Learning (PPAFL) framework is introduced to realize the decentralized model training for secure and resilient modern mobile robotic systems in 5G and beyond networks. Specifically, a reputation-aware coordination mechanism is designed and addressed to coordinate a group of smart devices dynamically into a virtual cluster, in which the asynchronous model aggregation is conducted in a decentralized P2P manner. A secret sharing based communication mechanism is developed to ensure an encrypted P2P FL process, while a Secure Stochastic Gradient Descent (SSGD) scheme is integrated with an Autoencoder and a Gaussian mechanism is developed to ensure an anonymized local model update, communicating within a few neighboring clients. The case study based experiment and evaluation in three different application scenarios demonstrate that the PPAFL can effectively improve the security and resilience issues compared with the traditional centralized approaches for smart mobile robotic applications in 5G and beyond networks.

Nowadays, the concept of Internet of Everything (IoE) is becoming a hotly discussed topic, which is playing an increasingly indispensable role in modern intelligent applications. These applications are known for their real-time requirements under limited network and computing resources, thus it becomes a highly demanding task to transform and compute tremendous amount of raw data in a cloud center. The edge-cloud computing infrastructure allows a large amount of data to be processed on nearby edge nodes and then only the extracted and encrypted key features are transmitted to the data center. This offers the potential to achieve an end-edge-cloud-based big data intelligence for IoE in a typical two-stage data processing scheme, while satisfying a data security constraint. In this study, a deep-reinforcement-learning-enhanced two-stage scheduling (DRL-TSS) model is proposed to address the NP-hard problem in terms of operation complexity in end-edge-cloud Internet of Things systems, which is able to allocate computing resources within an edge-enabled infrastructure to ensure computing task to be completed with minimum cost. A presorting scheme based on Johnson's rule is developed and applied to preprocess the two-stage tasks on multiple executors, and a DRL mechanism is developed to minimize the overall makespan based on a newly designed instant reward that takes into account the maximal utilization of each executor in edge-enabled two-stage scheduling. The performance of our method is evaluated and compared with three existing scheduling techniques, and experimental results demonstrate the ability of our proposed algorithm in achieving better learning efficiency and scheduling performance with a 1.1-approximation to the targeted optimal IoE applications.

The Internet of Things (IoT) has become a focus of information infrastructure development in recent years. The smart blockchain can provide various solutions for trust, security, and privacy (TSP) challenges to protect IoT data, and smart contracts are the foundation of blockchain intelligence, and greatly enhance the ability of smart blockchain to solve TSP problems. So the security of smart contracts must be addressed. We propose an efficient smart contract vulnerability detector to improve the safety of smart contracts. It comprises a graph extraction method and a complete vulnerability detection process. The graph extraction method consists of vulnerability pattern extraction and a graph generation process. The vulnerability detection process first uses the approximate graph matching algorithm to select representative SCGraphs from the dataset to build vulnerability SCGraph libraries. Secondly, determine whether the contract contains vulnerabilities by calculating the similarity between the SCGraphs generated from the contracts to be detected and the SCGraphs in the vulnerability library. Experiments show that our approach achieves an inspiring high detection rate and is the fastest among existing vulnerability detection tools, which indicates that it can provide good vulnerability detection for smart contracts.

The impact of Internet of Things (IoT) has become increasingly significant in smart manufacturing, while deep generative model (DGM) is viewed as a promising learning technique to work with large amount of continuously generated industrial Big Data in facilitating modern industrial applications. However, it is still challenging to handle the imbalanced data when using conventional Generative Adversarial Network (GAN) based learning strategies. In this article, we propose a distribution bias aware collaborative GAN (DB-CGAN) model for imbalanced deep learning in industrial IoT, especially to solve limitations caused by distribution bias issue between the generated data and original data, via a more robust data augmentation. An integrated data augmentation framework is constructed by introducing a complementary classifier into the basic GAN model. Specifically, a conditional generator with random labels is designed and trained adversarially with the classifier to effectively enhance augmentation of the number of data samples in minority classes, while a weight sharing scheme is newly designed between two separated feature extractors, enabling the collaborative adversarial training among generator, discriminator, and classifier. An augmentation algorithm is then developed for intelligent anomaly detection in imbalanced learning, which can significantly improve the classification accuracy based on the correction of distribution bias using the rebalanced data. Compared with five baseline methods, experiment evaluations based on two real-world imbalanced datasets demonstrate the outstanding performance of our proposed model in tackling the distribution bias issue for multiclass classification in imbalanced learning for industrial IoT applications.

Eyelid malignant melanoma (MM) is a rare disease with high mortality. Accurate diagnosis of such disease is important but challenging. In clinical practice, the diagnosis of MM is currently performed manually by pathologists, which is subjective and biased. Since the heavy manual annotation workload, most pathological whole slide image (WSI) datasets are only partially labeled (without region annotations), which cannot be directly used in supervised deep learning. For these reasons, it is of great practical significance to design a laborsaving and high data utilization diagnosis method. In this paper, a self-supervised learning (SSL) based framework for automatically detecting eyelid MM is proposed. The framework consists of a self-supervised model for detecting MM areas at the patch-level and a second model for classifying lesion types at the slide level. A squeeze-excitation (SE) attention structure and a feature-projection (FP) structure are integrated to boost learning on details of pathological images and improve model performance. In addition, this framework also provides visual heatmaps with high quality and reliability to highlight the likely areas of the lesion to assist the evaluation and diagnosis of the eyelid MM. Extensive experimental results on different datasets show that our proposed method outperforms other state-of-the-art SSL and fully supervised methods at both patch and slide levels when only a subset of WSIs are annotated. It should be noted that our method is even comparable to supervised methods when all WSIs are fully annotated. To the best of our knowledge, our work is the first SSL method for automatic diagnosis of MM at the eyelid and has a great potential impact on reducing the workload of human annotations in clinical practice.

Mobile crowdsensing (MCS) is a cost-effective paradigm for gathering real-time and location-related urban sensing data. To complete MCS tasks, MCS platform needs to exploit the trajectory of participants (vehicles or individuals, etc.) for effectively choosing participants. On one hand, the existing works usually assume that platform has possessed the abundant historical movement trajectory for participant selection, or can accurately predict the movement of participant before selection, but this assumption is impractical for many MCS applications, for some candidates have just arrived without sufficient mobility profiles, so-called trajectory from-scratch, or cold-trajectory issue. On the other hand, most of works only considers the coverage ratio of the sensing area, while some hotspots should be sensed frequently, so-called coverage degree of hotspots. To solve the issue, this paper proposes a reinforcement learning (RL) based, i.e., an improved Q-learning based online participant selection scheme to incorporate both coverage ratio and degree, PSARE. First, to solve the explosion of state-value table in traditional tabular Q-learning, an improved two-level Q-learning method is proposed to select participants in online way so as to achieve high long-term return. Specifically, in each selection round, PSARE dynamically compresses all the real participants (RPs) into several virtual participants (VPs) using the available historical trajectories of RPs, and the VP-based state-value table is constructed and constantly updated (i.e., the first level). Then, after selecting the VP through looking up the table, PARSE chooses the RP with the largest expected reward in this VP using epsilon-greedy way to balance the effect of exploration and exploitation (i.e., the second level). Moreover, the reward function is designed to measure the MCS coverage quality, including both coverage degree of hotspots and coverage ratio of target area. Thorough experiments on real-world mobility data set demonstrate that PSARE outperforms than other RL based online participant selection schemes (including deep Q-learning network) and traditional offline selection methods.

As the specific incarnation of cyber-physical-social systems, in deregulated electricity market, the market gaming behaviors may have significantly affected the costs of electricity delivered to the market. Especially, from the supply side, the primary goal of power generating companies (PGCs) is to develop strategic biddings to maximize their profits in long-term trading, when facing intrinsic uncertainty. Typically, in such repeated and dynamic settings, one fundamental challenge is that, any PGC neither has prior knowledge about all unknown opponents' incentives, nor observes their strategies and obtained profits. Especially, the common setting is that, once the bidding auction has occurred, the PGC only observes the market clearing price (MCP) at each round, and winning or losing status. While it is typical to assume some perfect or bounded rationality model of the PGCs, their real behaviors do not follow such assumptions due to lack of complete information, computational intractability, or lack of perfect execution, etc. We formulate the problem of sequentially optimizing any PGC's bids with an adversarial multi-armed bandit (MAB) model. Specifically, at each round, a PGC chooses to play against all other opponents from an infinite set of possible strategies that are split into continuous intervals by sequentially occurred MCPs. Then at the end of each round, the PGC observes the outcome of the auction and updates its estimation on the expected bid's fitness for each interval (i.e., how much the expected profit of the interval could be achieved), and selects the bid for the next round using the proposed algorithm Exp3C (i.e., exponential-weight for exploration and exploitation with continuous value). The experimental results based on real dataset demonstrate that Exp3C performs better than other heuristic schemes including pure greedy, ϵ-greedy and MCP predication based bidding schemes. Moreover, we theoretically prove the upper bound of average Exp3C regret per round follows O(2/√T), where T is the number of total rounds. In summary, the proposed Exp3C has two distinguished advantages. First it is distributed, since its decisions uniquely depend on its past decisions and profits. Second, it is rational, since a PGC is given guarantees on its own accumulated profit regardless of other PGCs' behaviors.

Accurate evaluation of the treatment result on X-ray images is a significant and challenging step in root canal therapy since the incorrect interpretation of the therapy results will hamper timely follow-up which is crucial to the patients' treatment outcome. Nowadays, the evaluation is performed in a manual manner, which is time-consuming, subjective, and error-prone. In this article, we aim to automate this process by leveraging the advances in computer vision and artificial intelligence, to provide an objective and accurate method for root canal therapy result assessment. A novel anatomy-guided multi-branch Transformer (AGMB-Transformer) network is proposed, which first extracts a set of anatomy features and then uses them to guide a multi-branch Transformer network for evaluation. Specifically, we design a polynomial curve fitting segmentation strategy with the help of landmark detection to extract the anatomy features. Moreover, a branch fusion module and a multi-branch structure including our progressive Transformer and Group Multi-Head Self-Attention (GMHSA) are designed to focus on both global and local features for an accurate diagnosis. To facilitate the research, we have collected a large-scale root canal therapy evaluation dataset with 245 root canal therapy X-ray images, and the experiment results show that our AGMB-Transformer can improve the diagnosis accuracy from 57.96% to 90.20% compared with the baseline network. The proposed AGMB-Transformer can achieve a highly accurate evaluation of root canal therapy. To our best knowledge, our work is the first to perform automatic root canal therapy evaluation and has important clinical value to reduce the workload of endodontists.

Recently, along with several technological advancements in cyber-physical systems, the revolution of Industry 4.0 has brought in an emerging concept named digital twin (DT), which shows its potential to break the barrier between the physical and cyber space in smart manufacturing. However, it is still difficult to analyze and estimate the real-time structural and environmental parameters in terms of their dynamic changes in digital twinning, especially when facing detection tasks of multiple small objects from a large-scale scene with complex contexts in modern manufacturing environments. In this article, we focus on a small object detection model for DT, aiming to realize the dynamic synchronization between a physical manufacturing system and its virtual representation. Three significant elements, including equipment, product, and operator, are considered as the basic environmental parameters to represent and estimate the dynamic characteristics and real-time changes in building a generic DT system of smart manufacturing workshop. A hybrid deep neural network model, based on the integration of MobileNetv2, YOLOv4, and Openpose, is constructed to identify the real-time status from physical manufacturing environment to virtual space. A learning algorithm is then developed to realize the efficient multitype small object detection based on the feature integration and fusion from both shallow and deep layers, in order to facilitate the modeling, monitoring, and optimizing of the whole manufacturing process in the DT system. Experiments and evaluations conducted in three different use cases demonstrate the effectiveness and usefulness of our proposed method, which can achieve a higher detection accuracy for DT in smart manufacturing.

With the increasing population of Industry 4.0, both AI and smart techniques have been applied and become hotly discussed topics in industrial cyber-physical systems (CPS). Intelligent anomaly detection for identifying cyber-physical attacks to guarantee the work efficiency and safety is still a challenging issue, especially when dealing with few labeled data for cyber-physical security protection. In this article, we propose a few-shot learning model with Siamese convolutional neural network (FSL-SCNN), to alleviate the over-fitting issue and enhance the accuracy for intelligent anomaly detection in industrial CPS. A Siamese CNN encoding network is constructed to measure distances of input samples based on their optimized feature representations. A robust cost function design including three specific losses is then proposed to enhance the efficiency of training process. An intelligent anomaly detection algorithm is developed finally. Experiment results based on a fully labeled public dataset and a few labeled dataset demonstrate that our proposed FSL-SCNN can significantly improve false alarm rate (FAR) and F1 scores when detecting intrusion signals for industrial CPS security protection.

With the increasing population of Industry 4.0, industrial big data (IBD) has become a hotly discussed topic in digital and intelligent industry field. The security problem existing in the signal processing on large scale of data stream is still a challenge issue in industrial internet of things, especially when dealing with the high-dimensional anomaly detection for intelligent industrial application. In this article, to mitigate the inconsistency between dimensionality reduction and feature retention in imbalanced IBD, we propose a variational long short-term memory (VLSTM) learning model for intelligent anomaly detection based on reconstructed feature representation. An encoder-decoder neural network associated with a variational reparameterization scheme is designed to learn the low-dimensional feature representation from high-dimensional raw data. Three loss functions are defined and quantified to constrain the reconstructed hidden variable into a more explicit and meaningful form. A lightweight estimation network is then fed with the refined feature representation to identify anomalies in IBD. Experiments using a public IBD dataset named UNSW-NB15 demonstrate that the proposed VLSTM model can efficiently cope with imbalance and high-dimensional issues, and significantly improve the accuracy and reduce the false rate in anomaly detection for IBD according to F1, area under curve (AUC), and false alarm rate (FAR).

Pneumothorax is a common pulmonary disease that can lead to dyspnea and can be life-threatening. X-ray examination is the main means to diagnose this disease. Computer-aided diagnosis of pneumothorax on chest X-ray, as a prerequisite for a timely cure, has been widely studied, but it is still not satisfactory to achieve highly accurate results. In this paper, an image classification algorithm based on the deep convolutional neural network (DCNN) is proposed for high-resolution medical image analysis of pneumothorax X-rays, which features a Network In Network (NIN) for cleaning the data, random histogram equalization data augmentation processing, and a DCNN. The experimental results indicate that the proposed method can effectively increase the correct diagnosis rate of pneumothorax, and the Area under Curve (AUC) of the test verified in the experiment is 0.9844 on ZJU-2 test data and 0.9906 on the ChestX-ray14, respectively. In addition, a large number of atmospheric pleura samples are visualized and analyzed based on the experimental results and in-depth learning characteristics of the algorithm. The analysis results verify the validity of feature extraction for the network. Combined with the results of these two aspects, the proposed X-ray image processing algorithm can effectively improve the classification accuracy of pneumothorax photographs.

Feature selection (FS) is one of the fundamental data processing techniques in various machine learning algorithms, especially for classification of healthcare data. However, it is a challenging issue due to the large search space. Binary Particle Swarm Optimization (BPSO) is an efficient evolutionary computation technique, and has been widely used in FS. In this paper, we proposed a Confidence-based and Cost-effective feature selection (CCFS) method using BPSO to improve the performance of healthcare data classification. Specifically, first, CCFS improves search effectiveness by developing a new updating mechanism that designs the feature confidence to explicitly take into account the fine-grained impact of each dimension in the particle on the classification performance. The feature confidence is composed of two measurements: the correlation between feature and categories, and historically selected frequency of each feature. Second, considering the fact that the acquisition costs of different features are naturally different, especially for medical data, and should be fully taken into account in practical applications, besides the classification performance, the feature cost and the feature reduction ratio are comprehensively incorporated into the design of fitness function. The proposed method has been verified in various UCI public datasets and compared with various benchmark schemes. The thoroughly experimental results show the effectiveness of the proposed method, in terms of accuracy and feature selection cost.

Coronavirus disease 2019 (COVID-19) is one of the most destructive pandemic after millennium, forcing the world to tackle a health crisis. Automated lung infections classification using chest X-ray (CXR) images could strengthen diagnostic capability when handling COVID-19. However, classifying COVID-19 from pneumonia cases using CXR image is a difficult task because of shared spatial characteristics, high feature variation and contrast diversity between cases. Moreover, massive data collection is impractical for a newly emerged disease, which limited the performance of data thirsty deep learning models. To address these challenges, Multiscale Attention Guided deep network with Soft Distance regularization (MAG-SD) is proposed to automatically classify COVID-19 from pneumonia CXR images. In MAG-SD, MA-Net is used to produce prediction vector and attention from multiscale feature maps. To improve the robustness of trained model and relieve the shortage of training data, attention guided augmentations along with a soft distance regularization are posed, which aims at generating meaningful augmentations and reduce noise. Our multiscale attention model achieves better classification performance on our pneumonia CXR image dataset. Plentiful experiments are proposed for MAG-SD which demonstrates its unique advantage in pneumonia classification over cutting-edge models. The code is available at https://github.com/JasonLeeGHub/MAG-SD.

Couplet is a unique art form in Chinese traditional culture. The development of deep neural network (DNN) technology makes it possible for computers to automatically generate couplets. Especially, Transformer is a DNN-based 'Encoder-Decoder' framework, and widely used in natural language processing (NLP). However, the existed Transformer mechanism cannot fully exploit the essential linguistic knowledge in Chinese, including the special format and requirements of Chinese couplets. Therefore, this article adapts the Transformer mechanism to generate meaningful Chinese couplets. Specifically, the contributions of our work are threefold. First, considering the fact that the words in the corresponding positions of the antecedent clause and the subsequent clause in a Chinese couplet always have same part-of-speech (pos, i.e., word class), pos information is intentionally added into the Transformer to improve the accuracy of the conceived couplet. Second, to deal with the large number of unregistered and low-frequency words in Chinese couplet, a specific unregistered/low-frequency word processing mechanism (UWP) is designed and combined with the Transformer model. Third, to further improve the coherence of couplets, we incorporate the polish mechanisms (PMs) into Transformer model. In terms of three evaluation criteria including bilingual evaluation understudy (BLEU), perplexity, and human evaluation, the experimental results demonstrate the effectiveness of our designed Chinese couplet generation system.

The smart grid is now increasingly dependent on smart devices to operate, which leaves space for cyber attacks. Especially, the intentionally designed false data injection attack (FDIA) can successfully bypass the traditional measurement residual-based bad data detection scheme. Considering that the smart grid data naturally contain linear and nonlinear components, inspired by parallel ensemble learning, especially by the stacking method, this article presents an effective two-level learner-based FDIA detection scheme using the Kalman filter and recurrent neural network (KFRNN). The first level includes two base learners, in which the Kalman filter is used for state prediction to fit linear data, and the recurrent neural network is used to fit the nonlinear data feature. The second-level learner uses the fully connected layer and backpropagation (BP) module to adaptively combine the results of two base learners. Then, through fitting Weibull distribution of the sum of square errors (SSEs) between the observed measurements and the predicted measurements, the dynamic threshold is obtained to judge whether FDIA occurs or not. Comprehensive simulation results show that our scheme has better performance than other neural network-based and ensemble learning-based FDIA detection schemes.

Scholarly big data, which is a large-scale collection of academic information, technical data, and collaboration relationships, has attracted increasing attentions, ranging from industries to academic communities. The widespread adoption of social computing paradigm has made it easier for researchers to join collaborative research activities and share academic data more extensively than ever before across the highly interlaced academic networks. In this study, we focus on the academic influence aware and multidimensional network analysis based on the integration of multi-source scholarly big data. Following three basic relations: Researcher-Researcher, Researcher-Article, and Article-Article, a set of measures is introduced and defined to quantify correlations in terms of activity-based collaboration relationship, specialty-aware connection, and topic-aware citation fitness among a series of academic entities (e.g., researchers and articles) within a constructed multidimensional network model. An improved Random Walk with Restart (RWR) based algorithm is developed, in which the time-varying academic influence is newly defined and measured in a certain social context, to provide researchers with research collaboration navigation for their future works. Experiments and evaluations are conducted to demonstrate the practicability and usefulness of our proposed method in scholarly big data analysis using DBLP and ResearchGate data.

Intelligent networking and big data analytics are two important pillars for the operation of systems. Edge computing is frequently used in smart IoT systems, particularly in those which cannot be served efficiently through cloud computing due to the limitations in bandwidth, latency and Internet connectivity. However, applications always generate a large amount of data, which are pre-programmed and predefined to run on the cloud or edge platform and can't be changed at run time. The applications may gain better performance if they synergistically run on the cloud and edge platform. In this study, a novel algorithm called Dynamic Switching Algorithm is proposed to ensure intelligent dynamics where all tasks are either offloaded on cloud or edge according to the system's real-time conditions. We further divide applications into four types based on their real-time requirements. Each type of application is set to a reasonable latency to make sure the system to have less processing time. The results demonstrate that our method outperforms two state-of-the-art methods, decreasing both the average delay and energy consumption of offloading by 8.17%~66.90% and 3.76%~78.60% respectively. The experimental evaluations show that the performance of the proposed method could effectively offload tasks in smart IoT systems.

Crowdsourcing has proved to be a splendid tool to aggregate the knowledge from a pool of individuals in order to perform abundant microtasks efficiently. Recently, with the explosive growth of online social network, Word of Mouth (WoM)-based crowdsourcing systems have emerged, in which besides conducting the tasks by themselves, participants simultaneously recruit other individuals through exploiting their social networks to help solve crowdsourced tasks. This crowdsourcing paradigm can greatly facilitate to grow the pool of crowdworkers. However, there exist two conflicting challenges in designing an effective WoM-based incentive mechanism: 1) sybil attack and 2) heterogeneous effect of participants. That is, intuitively, incentivizing (usually compensating for) common-ability individuals will inevitably stimulate the behavior of sybil attack (i.e., some individuals create multiple sybils, and split the total efforts into those sybils to expect more compensation). This paper proposes a novel biased contest-based crowdsourcing incentive mechanism through exploiting social networks (BciNet), aiming to balance those two conflicting objectives. BciNet is composed of two phases. First, based on spreading activation model, an enhanced geometric virtual point dissemination mechanism is able to provide sybil-proof property and accommodate the realistic social network structure. Second, based on participants' virtual points, a biased contest gives more reward to less able participants. Through carefully calibrating the bias factor, simulation results based on the real dataset show that BciNet can greatly improve the amount of participants' effort levels, and actually be robust against the sybil attack. In brief, for a practical incentive mechanism, the methodology to address conflicting goals is to put rational individuals into dilemma: to be sybil or not to be, it is the problem, i.e., the potential gain from the sybils in the second phase may be offset by the loss in the first phase.

Along with the advancement of several emerging computing paradigms and technologies, such as cloud computing, mobile computing, artificial intelligence, and big data, Internet of Things (IoT) technologies have been applied in a variety of fields. In particular, the Internet of Healthcare Things (IoHT) is becoming increasingly important in human activity recognition (HAR) due to the rapid development of wearable and mobile devices. In this article, we focus on the deep-learning-enhanced HAR in IoHT environments. A semisupervised deep learning framework is designed and built for more accurate HAR, which efficiently uses and analyzes the weakly labeled sensor data to train the classifier learning model. To better solve the problem of the inadequately labeled sample, an intelligent autolabeling scheme based on deep Q-network (DQN) is developed with a newly designed distance-based reward rule which can improve the learning efficiency in IoT environments. A multisensor based data fusion mechanism is then developed to seamlessly integrate the on-body sensor data, context sensor data, and personal profile data together, and a long short-term memory (LSTM)-based classification method is proposed to identify fine-grained patterns according to the high-level features contextually extracted from the sequential motion data. Finally, experiments and evaluations are conducted to demonstrate the usefulness and effectiveness of the proposed method using real-world data.

The crowdsourcing system is a distributed problem-solving platform, in which tasks are delivered to the crowd (i.e., crowdworkers) in the form of an open call. Usually, large-scale crowdsourcing systems contain abundant microtasks, and the overhead of a crowdworker spending on searching the appropriate task may be comparable to the cost of completing the task. Therefore, task recommendation is necessary. However, existing work ignores the dynamics in crowdsourcing system, i.e., new tasks continually arrive, which leads to the issues of task cold-start. To overcome the challenge of the new coming task recommendation, this article proposes a feature-based Bayesian task recommendation (BTR) scheme. The key idea to deal with the dynamics of the crowdsourcing system lies in that the BTR learns the latent factor of the task through the task features instead of task ID and then learns the user's preference according to their historical behaviors. Specifically, based on task features and the user's historical behavior records, BTR can not only timely provide crowdworkers with personalized task recommendations but also solve the task cold-start problem. The simulations based on the real crowdsourced data set demonstrate that BTR performs better than other typical schemes that target at recommending the newly arrived tasks to crowdworkers.

We are living in a cyber-physical-social environment with a variety of lifestyles and values. Living support has become important in such a diverse society. Owing to the ability to collect a large amount of personal data or life logs in the cyber-physical-social environment, it is now possible for us to provide living support based on personal data analysis. Moreover, analyzing such data can facilitate a deep understanding of an individual. In this study, we focus on the provision of cyber-enabled well-being oriented daily living support for an individual based on personal data analysis. Three categories of personal data are identified from an individual's daily life data. In this paper, we discuss the basic concept, model, and framework for well-being oriented personal data analysis in order to offer suggestions and advices to improve the living quality of an individual. Finally, we report a feasibility study with an application scenario by using personal and environmental data.

Deep neural network (DNN) learning has witnessed significant applications in various fields, especially for prediction and classification. Frequently, the data used for training are provided by crowdsourcing workers, and the training process may violate their privacy. A qualified prediction model should protect the data privacy in training and classification/prediction phases. To address this issue, we develop a differential privacy (DP)-enabled DNN learning framework, DNN-DP, that intentionally injects noise to the affine transformation of the input data features and provides DP protection for the crowdsourced sensitive training data. Specifically, we correspondingly estimate the importance of each feature related to target categories and follow the principle that less noise is injected into the more important feature to ensure the data utility of the model. Moreover, we design an adaptive coefficient for the added noise to accommodate the heterogeneous feature value ranges. Theoretical analysis proves that DNN-DP preserves ${\varepsilon }$ -differentially private in the computation. Moreover, the simulation based on the US Census data set demonstrates the superiority of our method in predictive accuracy compared with other existing privacy-aware machine learning methods.

The Internet of Things (IoT)-Cloud combines the IoT and cloud computing, which not only enhances the IoT's capability but also expands the scope of its applications. However, it exhibits significant security and efficiency problems that must be solved. Internal attacks account for a large fraction of the associated security problems, however, traditional security strategies are not capable of addressing these attacks effectively. Moreover, as repeated/similar service requirements become greater in number, the efficiency of IoT-Cloud services is seriously affected. In this paper, a novel architecture that integrates a trust evaluation mechanism and service template with a balance dynamics based on cloud and edge computing is proposed to overcome these problems. In this architecture, the edge network and the edge platform are designed in such a way as to reduce resource consumption and ensure the extensibility of trust evaluation mechanism, respectively. To improve the efficiency of IoT-Cloud services, the service parameter template is established in the cloud and the service parsing template is established in the edge platform. Moreover, the edge network can assist the edge platform in establishing service parsing templates based on the trust evaluation mechanism and meet special service requirements. The experimental results illustrate that this edge-based architecture can improve both the security and efficiency of IoT-Cloud systems.

While social computing related research has focused mostly on how to provide users with more precise and direct information, or on recommending new search methods to find requested information rapidly, the authors believe that network users themselves could be viewed as an important social resource. This study concentrates on analyzing potential and dynamic user correlations, based on topic-aware similarity and behavioral influence, which may help us to discover communities in social networking sites. The dynamically socialized user networking (DSUN) model is extended and refined to represent implicit and explicit user relationships in terms of topic-aware features and social behaviors. A set of measures is defined to describe and quantify interuser correlations, relating to social behaviors. Three types of ties are proposed to describe and discover communities according to influence-based user relationships. Results of the experiment with Twitter data are used to show the discovery of three types of communities, based on the presented model. Comparison with six different schemes and two existing methods demonstrates that the proposed method is effective in discovering influence-based communities. Finally, the scenario-based simulation of collective decision-making processes demonstrates the practicability of the proposed model and method in social interactive systems.

This paper focuses on seeking a new heuristic scheme for an influence maximization problem in social networks: how to economically select a subset of individuals (so-called seeds) to trigger a large cascade of further adoptions of a new behavior based on a contagion process. Most existing works on selection of seeds assumed that the constant number k seeds could be selected, irrespective of the intrinsic property of each individual's different susceptibility of being influenced (e.g., it may be costly to persuade some seeds to adopt a new behavior). In this paper, a price-performance-ratio inspired heuristic scheme, PPRank, is proposed, which investigates how to economically select seeds within a given budget and meanwhile try to maximize the diffusion process. Our paper's contributions are threefold. First, we explicitly characterize each user with two distinct factors: the susceptibility of being influenced (SI) and influential power (IP) representing the ability to actively influence others and formulate users' SIs and IPs according to their social relations, and then, a convex price-demand curve-based model is utilized to properly convert each user's SI into persuasion cost (PC) representing the cost used to successfully make the individual adopt a new behavior. Furthermore, a novel cost-effective selection scheme is proposed, which adopts both the price performance ratio (PC-IP ratio) and user's IP as an integrated selection criterion and meanwhile explicitly takes into account the overlapping effect; finally, simulations using both artificially generated and real-trace network data illustrate that, under the same budgets, PPRank can achieve larger diffusion range than other heuristic and brute-force greedy schemes without taking users' persuasion costs into account.

With the rapid popularity of social network applications and advanced digital devices, we have witnessed the explosive growth of multimedia big data in terms of both scale and variety over the last few years. Yet at the same time, security issues related to multimedia big data have arisen. An urgent demand has emerged for novel technologies that deal with copyright protection, multimedia forgery detection, and cybersecurity - especially for cyber-enabled multimedia applications. This special issue brings together research efforts in cybersecurity for cyber-enabled multimedia applications.

With the popularity of social networking services (SNSs) and the increase of users, individuals' social roles in a social network have become more and more important in terms of the recommendation of personalized services and the collective decision-making process. Usually, in an SNS system, active users may not represent the major opinions among the whole users, and most of the users' opinions may be multifarious. In this paper, we focus on analyzing and identifying users' dynamical social roles to facilitate the collective decision-making process. After introducing the social choice theory and an improved collective decision-making model, we present a three-layer model to analyze users' social roles in a hierarchical way and develop an integrated mechanism to utilize the identification of social roles to support the collective decision making. Based on a developed NetLogo-based tool, a case study for the course-offering determination with an application scenario is demonstrated to show the process of using users' social roles to support the collective decision making. The comparison experiment conducted between our method and the Delphi method shows the usefulness of our proposed method to help users achieve the decision consensus in a more efficient way.

The social Internet of Things and social cloud have the potential to move forward hand in hand, changing our social patterns and providing new ways to communicate and collaborate.

With the high development of social networks, collaborations in a socialized web-based learning environment has become increasing important, which means people can learn through interactions and collaborations in communities across social networks. In this study, in order to support the enhanced collaborative learning, two important factors, user behavior patterns and user correlations, are taken into account to facilitate the information and knowledge sharing in a task-oriented learning process. Following a hierarchical graph model for enhanced collaborative learning within a task-oriented learning process, which describes relations of learning actions, activities, sub-tasks and tasks in communities, the learning action pattern and Goal-driven Learning Group, as well as their formal definitions and related algorithms, are introduced to extract and analyze users' learning behaviors in both personal and cooperative ways. In addition, a User Networking Model, which is used to represent the dynamical user relationships, is proposed to calculate user correlations in accordance with their interactions in a social community. Based on these, an integrated mechanism is developed to utilize both user behavior patterns and user correlations for the recommendation of individualized learning actions. The system architecture is described finally, and the experiment results are presented and discussed to demonstrate the practicability and usefulness of our methods.

Sharing resources and information on the Internet has become an important activity for education. In distance learning, instructors can benefit from resources, also known as Learning Objects (LOs), to create plenteous materials for specific learning purposes. Our repository (called the MINE Registry) has been developed for storing and sharing learning objects, around 22,000 in total, in the past few years. To enhance reusability, one significant concept named Reusability Tree was implemented to trace the process of changes. Also, weighting and ranking metrics have been proposed to enhance the searchability in the repository. Following the successful implementation, this study goes further to investigate the relationships between LOs from a perspective of social networks. The LONET (Learning Object Network), as an extension of Reusability Tree, is newly proposed and constructed to clarify the vague reuse scenario in the past, and to summarize collaborative intelligence through past interactive usage experiences. We define a social structure in our repository based on past usage experiences from instructors, by proposing a set of metrics to evaluate the interdependency such as prerequisites and references. The structure identifies usage experiences and can be graphed in terms of implicit and explicit relations among learning objects. As a practical contribution, an adaptive algorithm is proposed to mine the social structure in our repository. The algorithm generates adaptive routes, based on past usage experiences, by computing possible interactive input, such as search criteria and feedback from instructors, and assists them in generating specific lectures.

In line with the popularity of the Internet and the development of search engine, users request information through web-based services. Although general-purpose searching such as one provided by Google is powerful, searching mechanism for specific purposes could rely on metadata. In distance learning (or e-learning), SCORM provides an efficient metadata definition for learning objects to be searched and shared. To facilitate searching in a federated repository, CORDRA provides a common architecture for discovering and sharing Learning Objects. We followed SCORM and CORDRA specifications to develop a registry system, called the MINE Registry, for storing and sharing 20,738 Learning Objects created in the past five years. As a contribution, we propose the concept of "Reusability Tree" to represent the relationships among relevant Learning Objects and enhance CORDRA. We further collect relevant information, while users are utilizing Learning Objects, such as citations and time period persisted. The feedbacks from the user community are also considered as critical elements for evaluating significance degree of Learning Objects. Through theses factors, we propose a mechanism to weight and rank Learning Objects in the MINE Registry, in addition to other external learning objects repositories. As a practical contribution, we provide a tool called "Search Guider" to assist users in finding relevant information in Learning Objects based on individual requirements.

Knowledge graph completion (KGC) aims at completing missing information in knowledge graphs (KGs). Most previous works work well in the transductive setting, but are not applicable in the inductive setting, i.e., test entities can be unseen during training. Recently proposed methods obtain inductive ability by learning logic rules from subgraphs. However, all these works only consider the structural information of subgraphs while ignoring the rich contextual semantic information underlying KGs, which tends to lead to a sub-optimal embedding result. Furthermore, they tend to perform poorly when the subgraphs are sparse. To address these problems, we propose a global and local Context-enhanced Embedding network, ConeE, which can fully utilize local and global contextual information to enhance embedding representations through the following two components. (1) The global context modeling module (GCMM) is a semi-parametric coarse-grained global semantic extractor, which can effectively extract global context-based semantic information via a BERT-based context encoder and a semantic fusion network (SFN), and adopts a novel contrastive learning-based sampling strategy to optimize semantic features. Furthermore, a scoring network is designed to evaluate the confidence of triplets from the perspective of both the triplet facts and the reasoning path to improve the accuracy of prediction. (2) The local context modeling module (LCMM) employs an interactive graph neural network (IGNN) to extract local topological features from subgraphs, and applies mutual information maximization (MIM) to subgraph modeling to capture more local features. Experiments on benchmark datasets show that ConeE significantly outperforms existing state-of-the-art methods.

Heterogeneous networks, like social and academic networks are widespread in the real world, characterized by diverse nodes and complex relationships. Influence maximization is a crucial research topic, in these networks, as it can help in comprehending the mechanisms of information propagation and diffusion. Effectively utilizing complex structural information poses a challenge in current research on influence maximization in heterogeneous information networks. As a solution to this problem, a heterogeneous network influence maximization algorithm based on the multi-scale propagation strength and repulsive force of propagation field is proposed. Firstly, based on the propagation field, we design a multi-scale propagation strength index for the propagation ability of nodes to achieve maximum coverage of influence propagation. Specifically, in the homogeneous structure, the homogeneous propagation strength describes the propagation ability of nodes. In the heterogeneous structure, the heterogeneous shallow propagation strength and the heterogeneous deep propagation strength are designed to exploit the local and global spreading ability of nodes using meta-paths and link prediction based on graph neural networks, respectively. Secondly, to ensure the minimum overlap in the propagation range of seed nodes, we designed the overlapping repulsive force between node pairs in the propagation field. Finally, considering the complexity of the propagation process of heterogeneous information networks, an independent cascade model based on meta-paths is proposed. Based on experiments conducted with several datasets, our algorithm outperforms baseline algorithms for solving influence maximization problem.

With the development of advanced embedded and communication systems, location information has become a crucial factor in supporting context-aware or location-aware intelligent services. Among these services, modern Intelligent Transportation System (ITS) has the strictest requirements for real-time, accurate, and private location data. In this study, a Spatial-Temporal Federated Transfer Learning (ST-FTL) framework is designed and introduced to achieve more precise cooperative positioning with multi-sensor data fusion while protecting the location data privacy in urban ITS. Specifically, a three-layer FTL architecture is constructed to enhance the prediction accuracy on GPS positioning errors for vehicles in different regions especially when facing missing local data in some specific scenarios (e.g., urban canyons), in which Transfer Learning (TL) is incorporated to optimize the initialization of global model with faster convergence but less communication cost in Federated Learning (FL). A multi-attribute based spatial-temporal clustering algorithm is developed to facilitate the finding of more appropriate source domains similar to the target domain in a density-based scheme, while a convolutional-gated unit is newly designed to further filter out the useless features and add new features based on the convolution operations and gating mechanism, resulting in more effective global model initialization and weight aggregation from cross-region selection. A multi-sensor data fusion model is built in local to improve the prediction accuracy on positioning errors, in which an improved time-aware asymmetric attention mechanism is involved to selectively adjust the weight importance of the incorporated GPS and Inertial Measurement Unit (IMU) data in a more targeted fusion process. Additionally, an improved Siamese network structure, which breaks the traditional dual network structure and only adopts one single network structure, is leveraged to realize the lightweight data augmentation based on the better utilization of historical similar data from local vehicles themselves, when suffering from insufficient training data. Experiments and evaluations based on two different public datasets demonstrate the outstanding performance of our proposed model and method in achieving superior prediction accuracy and faster convergence for cooperative positioning compared with other state-of-the-art positioning methods in urban ITS.

Stealthy FDIA (False Data Injection Attack) is a serious cyber threat that can modify state estimation of smart grid through maliciously altering the measurement data, but can't be detected by traditional bad data detection system in smart grid. There exist two weakpoints for numerous deep neural networks (DNNs) based data-driven schemes against FDIA. First, they mainly focus on detecting the presence of FDIA, but fail to localize the specific bus/nodes affected. Second, their performance is not sufficiently desirable under small attack, i.e., anomalies caused by the attack closely resemble normal data. To address the above issues, this paper proposes an effective locational FDIA detection framework based on data reconstruction, AT-GVAE, which seamlessly integrates the variational autoencoders (VAEs) and generative adversarial network paradigm. Specifically, our contributions are threefold. First, the proposed AT-GVAE framework is novelly composed of two main modules: the generative VAE_G and discriminative VAE_D that both play dual roles: reconstruct data from jointly learning distributions of data and latent feature space, and meanwhile play Minmax game with adversarial way. Second, multiple-layer gated recurrent units (GRUs) are utilized as the basic structure of the encoder and decoders in both VAEs, to characterize the temporal correlations of measurement data sequence. Additionally, the self-attention mechanism is used to enhance the expressive ability of GRU based VAEs. Then, the anomaly score for each busbar in the smart grid is determined by comparing the residual between the observed measurement and the outputs of VAE_G and VAE_D, enabling the localization of FDIA. Finally, thorough experiments on multiple power systems with series data of total 3456 measurements demonstrate that, in terms of multiple typical metrics, including false alarm rate vs. detection probability, AUC-ROC, and AUC-PR, our proposed framework outperforms the state-of-the-art GRU, VAE and adversarial training based FDIA detection schemes.

Accurate measurements of ophthalmic parameters and postoperative appearance prediction are essential for the diagnosis and treatment of many ophthalmic diseases. Nevertheless, it remains challenging due to (1) inconsistent ophthalmic image sampling standards, including ocular-camera distance, facial angle, and patient number, (2) complicated ocular morphology, such as subconjunctival hemorrhage, ocular movements, lighting effects, and morphological aging. It is difficult for a model to measure parameters and make predictions in variable sampling methods and morphology conditions. Therefore, the Global attention-based Ophthalmic Image Measurement and Postoperative Appearance Prediction System (GOMPS) is proposed, which quantifies ophthalmic image parameters to diagnose disease and simultaneously predict postoperative appearance of blepharoptosis. By perceiving the global structure of the ophthalmic image, GOMPS makes logical inference predictions of the sclera and cornea morphology, to overcome the above difficulties. Concretely, a global attention unit (GAU) and a novel global attention structure-aware network (GASA-Net) are designed to enhance GOMPS's global structure awareness ability to perform logical reasoning. Extensive experimental results on our collected ophthalmic dataset for diagnosis & prediction (OD2P) demonstrate that GOMPS surpasses the state-of-the-art methods in segmentation accuracy and achieves the current optimal performance in measurement and postoperative prediction under many clinical scenes.

Group behavior modeling is an important research topic in the field of social network analysis. Existing methods regarding this topic can only learn the static group preference, ignoring the multiple characteristics of the group behavior, so they cannot model the group behavior in a complete way. In this paper, we propose a Dynamic Multi-view Group Preference Learning (DMGPL) model for group behavior prediction in social networks. Firstly, an area-aware user dynamic preference extracting module is developed concerning user representation learning, which can integrate the dynamic user behavior preference and the corresponding group structural information into group members’ representations. Then, to simultaneously obtain stability, propensity, potentiality, and heterogeneity of the group behavior, Multi-view Group Preference Aggregation (MGA) is designed for group behavior modeling. In MGA, various aggregation strategies are used to capture different kinds of group behavior properties, which can achieve the effect of highlighting different properties in different scenarios. Moreover, considering the influence of the member's size on group behavior, we define the group scaling degree and perform group representation scaling to make MGA more flexible. Finally, extensive experiments are performed on two real-world datasets, and the results show that the accuracy of group behavior prediction by DMGPL improved by about 10% compared to the baseline models.

Abstract: Knowledge graph completion (KGC) aims at complementing missing entities and relations in a knowledge graph (KG). Popular KGC approaches based on KG embedding are typically limited to the transductive setting, i.e., all entities must be seen during training, which is impractical for real-world KGs where new entities are emerging daily. Recent inductive KG embedding approaches propose to train a neighborhood aggregator in conjunction with entity and relation embeddings, which helps to embed unseen entities via existing neighbors. However, existing methods do not fully take advantage of the structural information of neighbors and are unable to handle triplets involving unseen relations. In this paper, we work further and propose a novel and unified inductive KGC model, namely structure-aware attention network (SAAN), which can efficiently generate embeddings of unseen entities and relations by aggregating neighbors with structure-aware attention weights. Unlike conventional embedding-based attention methods, SAAN can naturally learn importance weights by modeling structural correlations between nodes in an embedding-independent manner and can be applied to any existing KG embedding model. Experimental results on both transductive and inductive KGC tasks show that our model significantly outperforms state-of-the-art methods. Graphical abstract: [Figure not available: see fulltext.]

Question answering over knowledge graph (KGQA) aims at answering natural language questions posed over knowledge graphs (KGs). Moreover, multi-hop KGQA requires reasoning across multiple triplets in KGs to get to the answer. Unfortunately, KGs often lack complete information and contain many missing links, which poses huge challenges for multi-hop KGQA. To address this, recent several approaches have introduced KG embedding techniques, which have shown good performance on the multi-hop KGQA task. However, these methods ignore the semantic correlations between paths and questions, and the reasoning process is not easily explained. Furthermore, traditional KG embedding methods consider only low-order pairwise relations and ignore the higher-order relations among entities, leading to a sub-optimal embedding result. To address these problems, we propose Hic-KGQA, a novel hypergraph and inference chain-based model for multi-hop KGQA. Specifically, Hic-KGQA first generates pre-trained entity embeddings with multiple semantics via a hypergraph-based KGC module (HKM). Then, an inference chain modeling module (ICMM) is designed to learn the importance of different inference chains for the question and encode the highest-ranked inference chain into an embedding representation. Finally, two scoring networks are used to evaluate the correlation between the candidate answers and the questions from the perspective of both the triplet facts and the reasoning process to obtain more accurate answers. Furthermore, the mutual information maximization (MIM) is innovatively implemented to capture richer common path features from similar cases to alleviate the missing path problem caused by KG incompleteness. Experiments show that Hic-KGQA significantly outperforms existing state-of-the-art methods and is explainable.

Accurate residential load forecasting (RLF) is of great significance for the decision-making and operation of modern power system. In literature, deep neural network (DNN) based RLF schemes have witnessed great development due to the advantage of automatically extracting features and capturing complex non-linear pattern in the presented data. However, most existed works separately exploit the historical data of a specific residential house to forecast its load. However, the electricity consumption behaviors among residential users are not independent, and implicitly have some correlations, which can be explicitly characterized and exploited to improve the accuracy of RLF forecasting. Inspired by this idea, through exploiting the multiple correlations among households, this paper proposes a novel residential load forecasting framework based on multiple correlation-temporal graph neural networks, RLF-MGNN. Specifically, the novelty of our work includes three aspects. First, multiple graphs are explicitly constructed to represent both linear and nonlinear correlations among temporal load series of households. That is, the synchronization graph is built to describe the degree of linear correlation between two households using Pearson correlation coefficient, which characterizes the similarity of their consumption behaviors, and the causality graph is built to describe their nonlinear correlation using transfer entropy, which characterizes the amount of directional information transfer from one time series to another, and models the mutual influence between households. Second, the multiple correlation-temporal graph convolutional networks (GCNs) are designed to forecast the residential users’ loads. In detail, at each timestep, latent features are first extracted by corresponding GCNs to embed multiple correlations among households, and then are sent to Long Short-Term Memory (LSTM) for further learning the latent temporal features. Finally, thorough experiments on real datasets demonstrate that our proposed RLF-MGNN outperforms the state-of-the-art independent DNN based schemes and other GNN based schemes.

Interactive Recommendation (IR) formulates the recommendation as a multi-step decision-making process which can actively utilize the individuals' feedback in multiple steps and optimize the long-term user benefit of recommendation. Deep Reinforcement Learning (DRL) has witnessed great application in IR for e-commerce. However, user cold-start problem impairs the learning process of the DRL-based recommendation scheme. Moreover, most existing DRL-based recommendations ignore user relationships or only consider the single-hop social relationships, which cannot fully utilize the social network. The fact that those schemes can not capture the multiple-hop social relationships among users in IR will result in a sub-optimal recommendation. To address the above issues, this paper proposes a Social Graph Neural network-based interactive Recommendation scheme (SGNR), which is a multiple-hop social relationships enhanced DRL framework. Within this framework, the multiple-hop social relationships among users are extracted from the social network via the graph neural network which can sufficiently take advantage of the social network to provide more personalized recommendations and effectively alleviate the user cold-start problem. The experimental results on two real-world datasets demonstrate that the proposed SGNR outperforms other state-of-the-art DRL-based methods that fail to consider social relationships or only consider single-hop social relationships.

The location technology of information sources in social networks is a major factor in exploring the means of information propagation. In this context, most existing methods ignore the direction of infected nodes and fail to make full use of the diffusion information, resulting in poor identification of source localization. To address such problems, a novel source location method based on infection potential energy is proposed. Its methodology consists of several steps: Firstly, a network reconstruction method is suggested, based on infection potential energy, to make the constructed information diffusion path more accurate. Afterwards, a network pruning method is identified to reduce the search space of candidate sources. Finally, considering the minor differences in the Gaussian density values of multiple nodes and the tendency of the rumor to spread, a distance centrality method is proposed. Experiments in real networks show that the suggested method achieves a better location performance.

To deal with the insufficient feedbacks and dynamics of individual arrival and item popularity in online recommender, collaborative multi-armed bandit (MAB) schemes intentionally utilize the explicitly known or implicitly inferred social relationships among individuals to collaboratively recommend. Especially, without assuming the social relationships among individuals given, the dynamic cluster of bandits simultaneously infers the relationships, and recommends items through using the inferred relationships in multi-round interactive steps. However, the existed clustering bandit algorithms have two weakpoints: first they either fix the number of clusters in advance, or assign two individuals into the same cluster if there exists a path between two users in graph structure, which may lead to the wrongly cluster users. Second, they usually exploit only the cluster’s accumulated parameters of cluster as the inferred preference of individual in the cluster, which can’t fully accurately learn individual’s latent preference. To address issues above, we propose new clustering MAB based online recommendation methods, ADCB and ADCB+, based on adaptively splitting and merging clusters, which incrementally enforce both user-level re-assignment and cluster-level re-adjustment in recommendation rounds to efficiently and effectively learn the individuals’ preferences and their clustering structure. Especially, the proposed ADCB+ method further exploits both the accumulated cluster preference parameters and each individual’s personalized feature through the adaptively weighting of the two influences according to the number of user interactions. The experiments on three real online rating datasets (i.e., MovieLens-2k, Delicious-2k, LastFM-2k) consistently show that, in terms of the cumulative reward over recommendation rounds, and the average Click-Through-Rate, our proposed ADCB and ADCB+ schemes outperform than some existing dynamic clustering based online recommendation methods.

In the mountainous areas of Japan, the weeds on the slopes of terraced rice paddies still need to be cut by the elderly manually. Therefore, more attention should be given to maintain proper postures while performing mowing actions (especially the pre-cutting actions) to reduce the risk of accidents. Given that complex mowing actions can be decomposed into different sub-actions, we proposed a joint angular calculation-based body movement analysis model based on the Hilbert–Huang transform to analyze the pre-cutting actions. We found that the two most important sub-actions were fast pre-cutting and slow pre-cutting. Based on field experiments, we analyzed the pre-cutting actions of workers with different experience levels and identified the factors that affected their falling risk (stability). The results showed differences and similarities in the actions’ frequency and amplitude in the sub-actions of workers with different mowing experience, confirmed the influence of body characteristics (body height, etc.) on body stability, and showed that workers should pay attention to their age and ankle part while mowing. The analysis results have identified factors for the mowing workers’ training and the development of equipment for use in complicated geographical conditions.

This paper explores the cause-and-effect relationships among a set of health indices using causal discovery. The data we used to analyze was obtained from wearable devices, Traditional Chinese Medicine (TCM) diagnosis, and self-assessment of subjects in an experiment. Firstly, three machine learning algorithms were employed to address the issue of excessive missing values in the integrated dataset, and the coherence of this improved data was validated by statistical test. The NOTEARS algorithm was then employed to assess the causal relationships within the overall population as well as within subgroups based on gender, physical activity levels, and sleep duration. The results demonstrated that the NOTEARS algorithm yielded interesting and plausible outcomes, suggesting the presence of causal connections between variables of wearable devices and TCM diagnosis, as well as daily lifestyle habits.

This paper explores human behavior in virtual networked communities, specifically individuals or groups’ potential and expressive capacity to respond to internal and external stimuli, with assortative matching as a typical example. A modeling approach based on Multi-Agent Reinforcement Learning (MARL) is proposed, adding a multi-head attention function to the A3C algorithm to enhance learning effectiveness. This approach simulates human behavior in certain scenarios through various environmental parameter settings and agent action strategies. In our experiment, reinforcement learning is employed to serve specific agents that learn from environment status and competitor behaviors, optimizing strategies to achieve better results. The simulation includes individual and group levels, displaying possible paths to forming competitive advantages. This modeling approach provides a means for further analysis of the evolutionary dynamics of human behavior, communities, and organizations in various socioeconomic issues.

Online learning provides flexibility and accessibility, however, lack of interaction and support leads to decreased satisfaction and brings feelings of loneliness. Studies explored ways to promote social interaction and collaboration among learners, but there remain challenges in addressing the issues of relying on asynchronous communication in online learning. In this study, we explored which pattern of scrolling comments is beneficial and effective for online learning support. Based on the i-Comments model, we design an experiment, in which three groups of subjects are randomly assigned to view the video only, the video with i-Comments pattern I and video with i-Comments patternII. During the learning process, the subjects’ concentration is detected by an eye-tracking device, while their peaceful level is monitored through the emWave system. After viewing the videos, the subjects are requested to take a quiz to test their comprehension, and a questionnaire is used to investigate cognitive load, fatigue, loneliness, and satisfaction. The results showed that i-Comments with more relevant knowledge, larger quantity, and appearance when viewing the third and last video were found to significantly improve learning comprehension, enhance learning satisfaction, keep a peaceful state, and alleviate the sense of loneliness in online learning.

The COVID-19 pandemic has been making big impact on mental and physical health of youth. Recent research shows that the COVID-19 pandemic has exacerbated existing mental health problems due to the unique combination of public health crises and social isolation. The objective of this study is to integrate and analyze various health data sources to improve health care for youth during the COVID-19 pandemic. The focus of the research is to merge self-assessment data from individuals, data obtained from wearable devices, and health data based on Traditional Chinese Medicine (TCM), utilizing machine learning techniques to gain a comprehensive perspective of youth health. The experiment results showed that the correlation between the TCM-based Health Score (TCMHS) in the TCM dimension and the Wearable Device Stress-based Health Score (WDSHS) in wearable devices was stronger than the correlation between the Self-assessed Subjective Health Score (SSHS) and the WDSHS. On the other hand, activity calorie consumption was the most important feature to both the SSHS and WDSHS while resting heart rate affected the TCMHS most.

It is important to study the evolution mechanism of group behavior for the prevention and control of harmful group behavior on social networks. Most researchers focus on the study of group behavior evolution in a deterministic environment, ignoring the influence of the uncertain environment formed by the randomness of behavioral decision-making, the incompleteness of public opinion information, and the instability of behavioral states. Therefore, this paper starts with the behavior decision-making process of individuals in an uncertain environment, defines the perception of public opinion, and builds a coevolution model of group behavior and opinion. First, considering the impact of public opinion information on group behavior evolution, this paper defines emotional resonance and topic popularity to quantify public opinion perception and designs an uncertain behavior decision-making model based on public opinion perception. Then, for the influence of the instability of the behavioral state on the opinion evolution process, this paper designs an opinion evolution model for perception based on the unstable behavioral state. Finally, we design the behavior state evolution mechanism based on the instability of the behavior state and fuse it with the behavior decision-making model and opinion evolution model to construct a coevolution model of group behavior and opinion. Compared with the experimental results of other models and real group behavior evolution, our model is effective.

The power of artificial intelligence of things (AIoT) stems from adapting machine learning (ML) and artificial intelligence (AI) models into abundant intelligent IoT fields, based on a large data stream with different formats, sizes, and timestamps generated by massive numbers of heterogeneous sensors. On the one hand, data acquisition is the fundamental basis for any AIoT systems, but data sensed by massive IoT devices may be noisy and even contain adversarial samples. On the other hand, ensuring the efficiency and robustness in data acquisition is vitally important for data-driven ML and AI. Recently, besides perceiving ability, the literature has witnessed great development of empowering things with learning and reasoning ability through deep learning models, including recurrent neural networks (RNNs) and/or convolutional neural network (CNNs). However, the existing works have one significant weakness: fail to explicitly leverage the geospatial implications and latent connections among sensors for high-quality data acquisition and quality control. Graphs are intrinsically suitable for representing the dependencies and inter-relationships between AIoT data sensing devices. Due to the ability of capturing the complex interactive relationships between nodes and producing high-level representations of the graph input, graph neural networks (GNNs) have exploded onto various ML and AI fields, to learn from graph-structured data. Our review covers the latest progresses in GNN for the fundamental atomic task of data acquisition in AIoT. Instead of surveying the abundant GNN schemes in vertically various IoT sensing applications, this paper systematically reviews the horizontal infrastructure that all AIoT fields should have, that is, AIoT data acquisition, based on GNN and other related emerging AI factors. Our contributions include the following aspects: Provide the latest progresses in GNN for the horizontal task of data acquisition in AIoT, propose the unified GNN pipeline based on encoder–decoder paradigm, and systematically categorize and summarize the emerging technologies helpful to address the issues in AIoT data acquisition, especially the noisy and adversarial data, and point out some future directions about GNN-based AIoT data acquisition.

As a distributed learning framework, Federated Learning (FL) allows different local learners/participants to collaboratively train a joint model without exposing their own local data, and offers a feasible solution to legally resolve data islands. However, among them, the data privacy and model security are two challenges. The former means that, if original data are used for trained FL models, various methods can be used to deduce the original data samples, thereby causing the leakage of data. The latter implies that unreliable/malicious participants may affect or destroy the joint FL model, through uploading wrong local model parameters. Therefore, this paper proposes a novel distributed FL training framework, namely LDP-Fed+, which takes into account differential privacy protection and model security defense. Specifically, firstly, a local perturbation module is added at the local learner side, which perturbs the original data of local learners through feature extraction, binary encoding and decoding, and random response. Then, through using the perturbed data, local neural network model is trained to obtain the network parameters that meet local differential protection, to effectively deal with model inversion attacks. Secondly, a security defense module is added on the server side, which uses the auxiliary model and differential index mechanism to select an appropriate number of local disturbance parameters for aggregation to enhance model security defense and deal with membership inference attacks. The experimental results show that, compared with other federated learning models based on differential privacy, LDP-Fed+ has stronger robustness for model security and higher accuracy for model training while ensuring strict privacy protection.

Background: In the post–COVID-19 pandemic era, many countries have launched apps to trace contacts of COVID-19 infections. Each contact-tracing app (CTA) faces a variety of issues owing to different national policies or technologies for tracing contacts. Objective: In this study, we aimed to investigate all the CTAs used to trace contacts in various countries worldwide, including the technology used by each CTA, the availability of knowledge about the CTA from official websites, the interoperability of CTAs in various countries, and the infection detection rates and policies of the specific country that launched the CTA, and to summarize the current problems of the apps based on the information collected. Methods: We investigated CTAs launched in all countries through Google, Google Scholar, and PubMed. We experimented with all apps that could be installed and compiled information about apps that could not be installed or used by consulting official websites and previous literature. We compared the information collected by us on CTAs with relevant previous literature to understand and analyze the data. Results: After screening 166 COVID-19 apps developed in 197 countries worldwide, we selected 98 (59%) apps from 95 (48.2%) countries, of which 63 (66.3%) apps were usable. The methods of contact tracing are divided into 3 main categories: Bluetooth, geolocation, and QR codes. At the technical level, CTAs face 3 major problems. First, the distance and time for Bluetooth- and geolocation-based CTAs to record contact are generally set to 2 meters and 15 minutes; however, this distance should be lengthened, and the time should be shortened for more infectious variants. Second, Bluetooth- or geolocation-based CTAs also face the problem of lack of accuracy. For example, individuals in 2 adjacent vehicles during traffic jams may be at a distance of ≤2 meters to make the CTA trace contact, but the 2 users may actually be separated by car doors, which could prevent transmission and infection. In addition, we investigated infection detection rates in 33 countries, 16 (48.5%) of which had significantly low infection detection rates, wherein CTAs could have lacked effectiveness in reducing virus propagation. Regarding policy, CTAs in most countries can only be used in their own countries and lack interoperability among other countries. In addition, 7 countries have already discontinued CTAs, but we believe that it was too early to discontinue them. Regarding user acceptance, 28.6% (28/98) of CTAs had no official source of information that could reduce user acceptance. Conclusions: We surveyed all CTAs worldwide, identified their technological policy and acceptance issues, and provided solutions for each of the issues we identified. This study aimed to provide useful guidance and suggestions for updating the existing CTAs and the subsequent development of new CTAs.

Nowadays, according to the report of world health organization, about 1.35 million people die in traffic accidents every year. Among them, because of it is difficult to predict in advance, the frequency of accidents caused by blind spots is relatively high, which means that it is necessary to design a system that can efficiently detect the target situations in which drivers in the area may have blind spots in their vision. Some existing methods, such as vehicle-mounted radar and pressure sensor are used for the blind spots' detection. However, they have some unresolved issues such as insufficient detection distance and limit excessive setting position. Therefore, to prevent misjudgment caused by insufficient information acquisition, this paper tries to present a multi-level object identification model based on face/object recognition and depth detection technologies via depth camera. Based on this, a new design of vision blind spot detection system is proposed by calculate the distance between human and vehicles via the depth camera. The proposed system has the function of determine whether vision blind spots exist and can quickly feedback the situation to drivers and pedestrians. This system will not only put forward a new vision blind spots determination method, but also can contribute to the research of public traffic accident prevention.

Owing to the aging of the rural population in the hilly and mountainous areas of Japan, mowing on narrow ridges and steep slopes is done manually by the elderly—individuals over 65 years of age. Studies have shown that many accidents that occurred during mowing were caused by workers’ unstable posture, especially when mowing on steep surfaces where there is a high risk of falling. It is necessary to analyze the body movements of mowing workers to elucidate the elements related to the risk of falls. Therefore, in this study, based on a high-precision motion-capture device and a series of experiments with elderly, skilled mowing workers, we focused on the movements of mowing. We sought to identify effective and safe mowing patterns and the factors that lead to the risk of falls. In various mowing styles, compared to the stride (S) and downward (D) mowing patterns, the basic (B) and moving (M) patterns were the most efficient; however, the risk of falls was also the highest among these patterns. While mowing, workers need to pay more attention to their arm strength and take appropriate measures to reduce the risk of falls according to their age and physique. The results can be used as data for the development of fall-detection systems and offer useful insights for the training of new mowing workers.

Online advertising click-through rate (CTR) prediction is aimed at predicting the probability of a user clicking an ad, and it has undergone considerable development in recent years. One of the hot topics in this area is the construction of feature interactions to facilitate accurate prediction. Factorization machine provides second-order feature interactions by linearly multiplying hidden feature factors. However, real-world data present a complex and nonlinear structure. Hence, second-order feature interactions are unable to represent cross information adequately. This drawback has been addressed using deep neural networks (DNNs), which enable high-order nonlinear feature interactions. However, DNN-based feature interactions cannot easily optimize deep structures because of the absence of cross information in the original features. In this study, we propose an effective CTR prediction algorithm called CAN, which explicitly exploits the benefits of attention mechanisms and DNN models. The attention mechanism is used to provide rich and expressive low-order feature interactions and facilitate the optimization of DNN-based predictors that implicitly incorporate high-order nonlinear feature interactions. The experiments using two real datasets demonstrate that our proposed CAN model performs better than other cross feature- and DNN-based predictors.

Nowadays, a variety of social media platforms acts as the major information portals, frequently updates the news and media pool and recommends them to dynamically arriving individuals, which requires real-time performance and immediate response to user feedback. Traditional offline/static recommendation approaches assume there exist large amount of historical interactive data, and build models from these data, which can neither deal with cold-start issue and nor fully incorporate individuals' feedbacks. Reinforcement learning based online recommendation system such as multi-armed bandit (MAB) and contextual MAB, attempts to maximize the long-term learning gains over time through exploring and exploiting the inherently interactive nature of online learning processes. However, most of work adopt the model-based paradigm, in which the feedback/reward function of individual is statically set as a specific format (e.g., linear in the features of the recommended contents and individual), which can only tailor to some specific learning objective and individual model. Without assuming the fixed reward model for each individual, this paper utilizes Gaussian process (GP) to characterize the expected feedback/reward of any recommended social media. Then, targeting at relatively accurate and noisy feedbacks of individual users, two MAB based algorithms are designed to select recommended social media to balance the effect of exploration and exploitation. The experimental results on real social media dataset Movielens demonstrate the effectiveness of our proposed methods compared with other model-based recommendation schemes.

Considering individuals can freely post messages on social media platforms, there is a large amount of unverified information, so-called rumor spreading on these platforms, which seriously affects users' experience and even disturbs social order. The application of Multi-Task Learning (MTL) in the field of rumor verification has witnessed great development, which improves rumor verification performance through jointly training the main task of rumor verification and the auxiliary task of stance classification. However, traditional MTL based rumor verification schemes can't adaptively weight different positions of data sequence to effectively represent the sequence, and then affect the verification performance. This paper proposes a novel rumor verification scheme for social media, MARV, through effectively exploiting the MTL and multi-head attention mechanism. Specifically, first, the shared LSTM layer in MARV is used to effectively process and represent the tweet sequences, and generate the high-level virtual features. Then, in the branch of rumor verification task, the multi-head attention layer is used to accurately learn the local dependencies in the high-level representations extracted from the shared layer. The experimental results on the PHEME and the RumourEval datasets demonstrate that our proposed MARV scheme is superior to other MTL based rumor verification schemes. Moreover, we also investigated the impact of differently placing attention module on the MTL based rumor verification.

Rapid growth of social network service (SNS) has drawn significant attention from the publics. Existing research indicated that emotional state and behavioral tendency of SNS users can be identified and predicted through sentiment analysis. However, it is found that not only the posts can express user’s emotions but the overall environmental conditions faced by the user may lead to the generation of different emotions based on the cognitive theory of emotion and observation. Therefore, it may lead to bias between the sentiment analysis results and the actual situation if only analyzing the post. This study targets to propose an extendable sentiment monitoring model which considers the actual environment of users in SNS. Through this model, the result of sentiment analysis is closer to reality. By analyzing the content of users’ continuous posts, the sentiment analysis can take into account the pre- and post-textual relationships. The classification result of external affecting sentiment factors by K-means is used as criteria for weighting method to adjust the results of sentiment analysis based on BERT. Finally, the time series analysis is used to predict sentiment tendency monitor sentiment changes. The experiment results show that the training and validation accuracy are 89.24% and 84.00%, respectively. By our weighting method to revise the BERT results, the F1 score is improved from 0.839 to 0.850.

In recent years, SNS has become a service that everyone uses. In this study, we analyze Twitter, one of the most popular SNSs, which allows users to post their daily events and feelings within 140 characters and is used by people all over the world. In this study, we investigate the relationship between SNS posts and latent approval needs. The linguistic features of tweets and their contextual features are analyzed using information such as the frequency of posts and the number of characters in tweets, and the degree of desire for approval is defined and quantified based on the results of the analysis of tweets. The experiment results show that the agreement between the naïve Bayes classifier and human ratings was about 60%. It is found that users with a high percentage of posts for approval desire tend to post less frequently and with a higher average number of characters. This indicates that it may be because these users post for approval desire when it is important or when they really want to say something.

Many of the texts posted on Twitter are broken sentences, and the translated sentences may not be accurate. An inaccurate translation may spoil the meaning of the original text and induce miscommunication between the poster and the reader who uses the machine translation. Since many sentences tweeted on Twitter contain emotional expressions, this study uses sentiment analysis to calculate and compare the sentiment scores of the original and translated sentences to investigate the change in sentiment before and after machine translation. As a result of using dictionaries to classify tweets before and after translation, it was found that the classification of positive sentences tended to be more likely the same before and after translation. In addition, the results of the sentiment analysis of “joy”, “like”, “relief” and “excitement” by machine learning showed that the sentiment of “joy” was particularly increased when translated from Japanese into English.

It is believed that the living habits and living environment have a close relationship with a person’s concentration. In this study, an experiment is designed to measure the degree of concentration for two subjects for 40 days. Living habits and environment data, as well as EEG (Electroencephalogram) data are collected and divided into six cases according to gender and concentration status, and then PCA (Principal Component Analysis) for each case is conducted. Furthermore, using the principal components identified by PCA, a regression model is constructed to analyze the relationship between lifestyle (living habits and environment) data and concentration indexes by EEG. The analysis results on all the data of the subjects suggested that the regression model has a certain degree of accuracy and there exists a significant relationship between concentration and the time and quality of sleep on the previous day, even if the concentration status on the following day is different.

With the rapid development of digitization in education, online learning has become an alternative and promising way to enable flexible learning scenarios. Especially after COVID-19, online learning draws more and more attention. However, online learning has inherent disadvantages and problems. One of them is that learners have difficulty in concentrating, and the other one is that they may feel lonely during lessons. In this study, a new model of on-screen individualized comments, namely i-Comments, is proposed for online learning support. In the proposed model, three elements, i.e., timing, content, and quantity of i-Comments, are defined respectively, which are used to help learners improve concentration and decrease the feeling of loneliness. Furthermore, an experiment is designed to evaluate the model and verify whether the proposed model can help learners gain a better learning experience and improve their learning effectiveness.

Most of previous research works of sentiment analysis on SNS mainly focused on polarity analysis to probe into user tendencies. However, human emotions are complex and changeable. It is difficult to use the results of traditional polarity analysis in the real-time application services. Although finer-grained sentiment analysis may provide more detailed results, it has the problem of ambiguity in the definition of features between emotions. In this study, we propose a partial redistribution method based on BERT to tackle this problem. It improves emotional confusion in sentiment analysis through the confusion matrix, and further uses binary classification models to re-train the data of the confused emotional group through the redistribution process. In addition, the model makes it possible to re-extract and define features for specific emotions. Finally, F1 score is used to judge whether each feature correction process exerts positive impact on the model. Experimental results demonstrate that our proposed approach is effective in improving emotional confusion issues in SNS sentiment analysis.

The development of interface designs can reduce extraneous processing for users and increase the effectiveness of multimedia presentations. In this study, we investigate cognitive load in multimedia presentations. First, we present a quantitative model to measure cognitive load in terms of information comprehension in which the pupil diameter variation is used as an indicator of cognitive load based on cognitive load theory. We design a verification experiment to measure the pupil diameter using an eye-tracker when different combinations of texts, audio-narrations, and images are presented to subjects. We further allow the subjects take a comprehension test on the presented information and analyze the relationship between cognitive load and the test score using the generalized linear mixed model (GLMM). Moreover, we obtain the subjective cognitive load via a pre-designed questionnaire taken after the experiment and compare these two types of cognitive loads in terms of the mean absolute error (MAE). The experiment results show that there is a gap between the objective cognitive load obtained via the pupillary response and the subjective cognitive load obtained via a questionnaire, and the presentation with an optimized combination of multimedia can enhance information comprehension while reducing cognitive load.

The direct release of medical image may face the dilemma: the privacy protection of medical images inevitably affects the visual quality of images. To balance medical image quality and privacy, this paper proposes a quality-aware and privacy-preserving medical image release scheme, QAPP, which effectively integrates the discrete cosine transform (DCT) with differential privacy (DP). Specifically, QAPP is composed of three phases. First, DCT is applied to each medical image to obtain its cosine coefficients matrix. Second, the original cosine coefficients matrix is compressed into k*k cosine coefficients matrix, which can retain the main features of each image. Third, the appropriate Laplace noise is injected into the formed k*k matrix to achieve differential privacy, and these noise-added coefficients are used to reconstruct the noise-added medical images through inverse DCT. Especially, considering there two error sources affecting the image quality in our work: the compression error caused by DCT, and the injected noise error caused by DP, Therefore, a selection function is proposed to determine the optimal compression dimension k, which can minimize the influence of these two errors to improve the visualization quality of the medical image. Subjective and objective image quality evaluation, and extensive experiments of image classification and segmentation using the real medical image dataset demonstrate that the proposed method QAPP can better balance medical image quality and privacy than other similar DP-based methods.

Peer assessment has become a primary solution to the challenge of evaluating a large number of students in Massive Open Online Courses (MOOCs). In peer assessment, all students need to evaluate a subset of other students’ assignments, and then these peer grades are aggregated to predict a final score for each student. Unfortunately, due to the lack of grading experience or the heterogeneous grading abilities, students may introduce unintentional deviations in the evaluation. This paper proposes and implements a semi-supervised peer assessment method (SSPA) that incorporates a small number of teacher’s gradings as ground truth, and uses them to externally calibrate the procedure of aggregating peer grades. Specifically, each student’s grading ability is directly (if students have common peer assessments with teacher) or indirectly (if students have no common peer assessments with teacher) measured with the grading similarity between the student and teacher. Then, SSPA utilizes the weighted aggregation of peer grades to infer the final score of each student. Based on both real dataset and synthetic datasets, the experimental results illustrate that SSPA performs better than the existing methods.

The spread of rumors has a major negative impact on social stability. Traditional rumor spreading models are mostly based on infectious disease models and do not consider the influence of individual differences and the network structure on rumor spreading. In this paper, we propose a rumor Fick-spreading model that integrates information decay in social networks. The dissemination of rumors in social networks is random and uncertain and is affected by the dissemination capabilities of individuals and the network environment. The rumor Fick-transition coefficient and Fick-transition gradient are defined to determine the influence of the individual transition capacity and the network environment on rumor propagation, respectively. The Fick-state transition probability is used to describe the probability of change of an individual's state. Moreover, an information decay function is defined to characterize the self-healing probability of individuals. According to the different roles and reactions of users during rumor dissemination, the user state and the rumor dissemination rules among users are refined, and the influence of the network structure on the rumor dissemination is ascertained. The experimental results demonstrate that the proposed model outperforms other rumor spread models.

Behavior is autonomous, convergent, and uncertain, which brings challenges to the modeling of social network behavior spread. In this article, we propose a behavior spread model based on group cohesion under uncertain environments. First, for behavioral convergence, we define group cohesion to quantify the convergent effects of group. Second, based on the game theory to model the autonomy of behavior, according to the characteristics of the game payoffs changing with time and the depth of spread, and integrating group cohesion, a dynamic game payoffs calculation method is designed. Finally, aiming at the uncertainty of behavior, a group behavior spread model based on random utility theory is established. Experiments on multiple real social network behavior spread datasets demonstrate the effectiveness of the proposed model in modeling and predicting behavior spread processes under uncertain environments.

In recent years, personal health data can be collected via wearable devices and sensors and used for healthcare services improvement through data sharing. To share sensitive personal health data securely, many frameworks and approaches using blockchain-based systems have been proposed. However, the issue of letting individuals control and manage their data with privacy-preserving is still to be solved. In this paper, we propose a new model of Individual-Initiated Auditable Access Control (IIAAC) enabled with blockchain, CP-ABE (Ciphertext-Policy Attribute-Based Encryption) and IPFS (Inter-Planetary File System) for privacy-preserved data sharing. We describe the system architecture, its main components for our proposed model, and the protocols to make blockchain and IPFS interoperate with each other via a client application, including key generation, data publication, and data retrieval. We further build an experiment environment to evaluate the performance of our proposed model and architecture. Experiment results demonstrate the feasibility of our proposed model and system architecture for privacy-preserved data sharing.

Previous studies have shown that about 90% of traffic accidents are due to human error, which means that human factors may affect a driver's braking behaviors and thus their driving safety, especially when the driver makes a braking motion. However, most studies have mounted sensors on the brake pad, ignoring to some extent an analysis of the driver's behavior before the brake pad is pressed (pre-braking). Therefore, to determine the effect of different human factors on drivers' pre-braking behaviors, this study focused on analyzing drivers' local joints (knee, ankle, and toe) by a motion capture device. A Hilbert–Huang Transform (HHT)-based local human body movement analysis method was used to decompose the realistic complex pre-braking actions into sub-actions such as intrinsic mode functions (IMF1, IMF2, etc.). Analysis of the results showed that IMF1 is a common and necessary action when pre-braking for all drivers, and IMF2 may be the safety assurance action that allows right-foot transverse movement at the beginning part of the pre-braking process. We also found that the experienced, male, and Phys.50 groups may have consistent characteristics in the HHT scheme, which could mean that such drivers would have better performance and efficiency during the pre-braking process. The results of this study will be useful in decomposing and discerning the specific actions that lead to accidents, providing insights into driver training for novice drivers, and guiding the construction of daily automated driver assistance or accident prevention systems (advanced driver assistance systems, ADASs).

Location-related data are an important subset of personal data. An individual may have a positive or negative feeling for a specific place, which is important for personal data analysis. There are many studies on sentiment analysis within text data, such as tweets, but few studies have been conducted specifically on an individual's feelings regarding locations. In this study, we focus on measuring and quantifying an individual's feelings for a place using three representative methods in sentiment analysis: emotion dictionary, personalized dictionary, and Bayesian classification. We design an experiment to evaluate these methods using tweet data including locations and an individual's emotional changes with regard to these locations before entering, after exiting, and in a location. Three sets of emotion scores are obtained and normalized. Furthermore, we set four protocols and use statistical methods to compare these emotion scores with the subjective emotion scores provided by the user whose tweets are used in the experiment. Experimental results show that Bayesian classification performs the best in measuring and quantifying an individual's feelings for a place.

Due to the aging of rural population, in Japan's terraced field areas, many elderly people have to do the mowing works of paddy levees manually. However, it takes the high risk of falling when mowing on slops, and many accidents are caused by the unstable postures. The analysis of their body motion becomes necessary. Based on our previous study, the personal factors could be important on affecting the mowing behaviors. Therefore, in this paper, we focus more on the mowing workers' personal factors (such as age and physique) and try to analyze the effect on their falling risk via a high precision motion capture device. A set of experiments was conducted in September 2020, and four new mowing workers over 60 years old were invited to participate in the experiments. Their joints angles data and related personal factor data were used to analyze the factors that have the greatest influence on body stability during slop mowing. The results of stepwise regression showed that the mowing workers' age and physique have significant effects on the stability of their body motion. The results of this study can clarify the factors that prevent accidents in the fall detection systems and offer useful insights for the training of new mowing workers.

In smart grid regarded as specific embodying of cyber-physical-social system, load forecasting, especially short-term load forecasting for individual electric customers plays an increasingly role in planning and operation of smart power system. Considering the similarity of electricity consumption between users, inspired by multi-task learning, the article puts forward an effective residential load forecasting based on multi-task learning model. In detail, the K-means clustering technology and Pearson correlation coefficient are used to select two similar users. Then these two user's load data are merged as input, the bidirectional long short-term memory network is used as a sharing layer to fully capture the relationship between the data of the two users, and then two fully-connection task-specific output layers are respectively built. Based on real datasets, the proposed scheme is thoroughly compared with several typical deep learning based load forecasting schemes. Experiments show that proposed multi-task learning scheme improves the prediction accuracy compared with the existing deep learning prediction scheme.

With the rapid development of sensors and IoT technology, personal health data can be collected and stored by various wearable devices and utilized for healthcare. To share and use sensitive health data securely and efficiently, a variety of solutions based on blockchain have been proposed and developed. However, there are still many issues to be solved, such as how to let individuals control and manage their own data, and how to make all data accesses strictly auditable. In this paper, we present a new model of Individual-Initiated Auditable Access Control (IIAAC) enabled with blockchain, CP-ABE (Ciphertext-Policy Attribute-Based Encryption) and IPFS (InterPlanetary File System). After introducing scenarios for sharing and use of health data, we define the design requirements for a blockchain-based system and describe the basic system architecture. We discuss the detailed procedures in IIAAC, including CP-ABE key generation, data publication and data retrieval. We further compare this study with related work in terms of functions and features.

Traditional social community discovery methods concentrate mainly on static social networks, but the analysis of dynamic networks is a prerequisite for real-time and personalized social services. Through the study of community changes, the community structure in a dynamic network can be tracked over time, which helps in the mining of dynamic network information. In this paper, we propose a method of tracking dynamic community evolution that is based on resistance distance. Specifically, we model the time-varying features of dynamic networks using the convergence of a resistance-based distance. In our model, the heterogeneity of neighboring nodes can be obtained in the local topology of nodes by analyzing the resistance distance between nodes. We design a community discovery algorithm that essentially discovers community structures on dynamic networks by identifying the so-called core node. During the process of community evolution analysis, both the dynamic contribution of ordinary nodes and core nodes in each community are considered. In addition, to avoid the inclusion of spurious communities in the community structure, we define the notion of noise community and account for it in our algorithm. Experimental results show that the method proposed in this paper can yield better accuracy than other existing methods.

With the surge in popularity of wearable devices, collection of personal health data has become quite easy. Many studies have been conducted using health data to estimate the onset and progression of illness. However, life habits may vary among individuals. By analyzing the life cycle from health-related data, conventional studies may be improved. This study proposes a new approach to improving diagnosis estimation by considering the life cycle analyzed from health-related data. The periodic span of the life cycle is estimated via autocorrelation analysis. In the range of the periodic span, dimension reduction for health data is performed by principal component analysis, and health features are extracted and used for diagnosis estimation. In our experiment, we used personal health data and pulse diagnosis data collected by a traditional Chinese medicine doctor. Using six multi-label classification methods, we verified that a combination of pulse and health features could improve the accuracy of diagnosis estimation compared with that using only pulse features.

Background and objectives: Infection with human papillomavirus (HPV) is the main cause of cervical cancer, and vaccination is an effective method to prevent HPV infection. In Japan, adverse reactions were reported in some HPV-vaccinated people in March 2013, and while Japan’s Ministry of Health, Labor, and Welfare withdrew active recommendation of the vaccine in June 2013, the social movement to refuse vaccination has continued. The Ministry of Health, Labor, and Welfare (MHLW) has devised a plan to accurately disseminate information that promotes vaccination, but less than 1% of the eligible population was vaccinated, and the number has not increased. Besides, inaccurate information about health information can disseminate rapidly on social networks. Social networking services (SNS), mainly used by young people, can be used by the public to obtain medical information. However, according to the World Health Organization (WHO), SNSs are prone to spreading inauthentic and misleading information when it comes to information related to health and medical care. “Infodemic” is defined as a situation in which unidentified and false information is widely disseminated on SNS, causing WHO to issue international alerts. This study aimed to organize information about HPV vaccination disseminated on SNS in Japan. Methods: We extracted 208 tweets with the keyword “HPV vaccine” posted in Japan between April 1, 2014, and September 30, 2017. The extracted tweets included data points such as ID, language, posting date and time, and latitude and longitude. The location information of the senders was obtained from the latitude and longitude, and the tweets were organized by prefecture, city, town, village, and ID. Then, we evaluated whether the information at the URLs was accurate by examining retweets, likes, and the number of comments on the tweet. Python version 3.7.7 was used to extract the tweets. Results: The results of classification of the tweets by prefecture are as follows: the Hokkaido prefecture accounted for four tweets; the northeast, six tweets; southern Kanto, 123 tweets; northern Kanto-Koshin, six tweets; Hokuriku, five tweets; Tokai, 35 tweets; Kinki, 10 tweets; Chugoku 4 tweets; Shikoku, three tweets; and Kyushu, nine tweets. A total of 93 users posted tweets; four users posted five or more tweets; 14 users posted 2–4 tweets, and 75 users posted one tweet. In particular, 66 tweets in Kanagawa prefecture, 14 tweets in Shizuoka prefecture, and two tweets in Tokyo were posted from the same ID. Regarding the type of tweet, there were 109 tweets, 65 retweets, and 34 replies. There were 137 tweets with and 71 tweets without URLs. When organized by the linked URL, 50 posts linked to a blog, 46 posts linked to a news item, seven posts linked to Facebook, five posts linked to a government agency homepage, four posts linked to YouTube, three posts linked to the home page of the City Council rep, two posts linked to a medical site, and 20 posts linked to other sources that could not be categorized. In terms of the authenticity of the posts, 25 tweets were judged as “accurate,” 14 were judged to be “inaccurate,” and 16 were judged as “unknown.” We classified the posts as follows; “accurate” for those that contained accurate information and “inaccurate” for those that contained inaccurate information. Discussion: The distribution of tweets and the uneven distribution of the users suggest that few people spread information about the HPV Vaccine on Twitter in Japan. Regarding the content, more than half of the tweets could not be judged as accurate or inaccurate because the verification results regarding adverse reactions of the HPV vaccine were not published at the time of sending, and in news and blog articles, personal opinions were stated rather than authentic medical information. In this study, we clarified the characteristics of tweets regarding HPV vaccination in Japan and the status of transmission. In the future, it will be necessary to change the keywords and time periods for which tweets need to be extracted, and the data set used for the analysis will need to be compared and examined.

In the era of big data, information about the same objects can be gathered and accumulated from multiple sources (i.e., crowdworkers) through so-called crowdsensing. Especially, in the indoor positioning system using Bluetooth fingerprints, multiple crowdworkers are required to collect the information of Bluetooth beacons at the reference points and the corresponding received signal strength indicators (RSSI). Due to the unknown proper and bias of each crowdworker, it is challenging to appropriately estimate the reliability of each worker/source and truthfully aggregate data. Moreover, the collected data possesses two properties: they follow long-tail, where most of the data is gathered by a few sources, i.e., abundant crowdworkers only provide small amount; they have time-sequential feature: the truth about the crowdsensing tasks smoothly evolve with time. In response to the above problem and data features, this paper proposes an accurate data aggregation mechanism incorporating sequential long-tail characteristics, DAST. Specifically, we infer each source's credibility based on the estimated confidence interval using the amount of data historically provided by the source. Meanwhile, in order to capture the sequential characteristics of the data, the accumulated data in previous period is used as a virtual source to obtain the new aggregated value for the current period. Thorough simulations using artificial data and real data demonstrate that the performance of DAST is superior to the existed schemes including Confidence-Aware Truth Discovery (CATD), Precision-Recall (PrecRec) and Dynamic Truth Discovery (DynaTD) in terms of the mean absolute error (MAE) and the root mean square error (RMSE).

To achieve an accurate assessment of root canal therapy, a fundamental step is to perform tooth root segmentation on oral X-ray images, in that the position of tooth root boundary is significant anatomy information in root canal therapy evaluation. However, the fuzzy boundary makes the tooth root segmentation very challenging. In this paper, we propose a novel end-to-end U-Net like Group Transformer Network (GT U-Net) for the tooth root segmentation. The proposed network retains the essential structure of U-Net but each of the encoders and decoders is replaced by a group Transformer, which significantly reduces the computational cost of traditional Transformer architectures by using the grouping structure and the bottleneck structure. In addition, the proposed GT U-Net is composed of a hybrid structure of convolution and Transformer, which makes it independent of pre-training weights. For optimization, we also propose a shape-sensitive Fourier Descriptor (FD) loss function to make use of shape prior knowledge. Experimental results show that our proposed network achieves the state-of-the-art performance on our collected tooth root segmentation dataset and the public retina dataset DRIVE. Code has been released at https://github.com/Kent0n-Li/GT-U-Net.

Recently, on one hand, human activity recognition (HAR) has witnessed great application on portable smart devices (e.g., smart phones and wearables, etc.) as they are widely used around the world. On the other hand, HAR methods based on deep learning have attracted much attention, for they possess excellent performance due to their strength on extracting virtual features automatically and hierarchically. However, to establish a personalized deep learning based HAR scheme based on smart devices, insufficient records from target users and heavy computation cost on training from scratch are two challenges. Considering that, in transfer learning, the knowledge learnt in the source domain could be appropriately transferred to help accomplish tasks in the target domain, this paper proposes a personalized HAR scheme through exploiting virtual feature adaptation based on transfer learning (i.e., VFAT) to achieve high recognition accuracy with low computation time. VFAT is composed of pre-Training phase on sufficient labeled records in source-domain, and adaption phase on target-domain that uses the few labeled records available. Specifically, VFAT scheme pre-Trains the LSTM-based feature extraction component in the pre-Training phase and then introduces domain loss in the adaptation phase to minimize the similarity between target-domain virtual features and source-domain activity patterns (i.e., virtual features averaged by activity labels). The HAR scheme applied to the MotionSense dataset and results demonstrate the effectiveness of our proposed VFAT scheme. Moreover, we also investigate the impact of domain division on the performance of transfer learning based HAR.

Front-end frameworks are in increasing demand in web application development. However, it is difficult to compare them manually because of their rapid evolution and big variety. Prior research has revealed several indicators that developers consider important when selecting a framework. In this study, we propose and develop a system that assists developers in the selection process of a front-end framework, which collects data from repository and user community, such as GitHub and other sources, and quantifies a set of indicators. This system is built as a web application, and users can specify the importance of an indicator by adjusting the weight for each indicator. As a result of semi-structured interviews with front-end developers after using our system based on practical scenarios, we found that the proposed system is effective for narrowing down the framework and has practicality.

The unprecedented development of the Internet of Things (IoT), artificial intelligence (AI), and Big Data has stimulated a boom of social networks such as Twitter, WeChat, Facebook, etc., generating a huge amount of social data that are worth further analysis. Social computing has an important focus on mining the deep relationships between social organizations, networks, and media. The increasing volumes and complexities make big social data mining more and more difficult. Hybrid Human-Artificial Intelligence (H-AI) is an approach combining both human intelligence and AI, so as to handle demanding problems in a harmonious way. By adopting H-AI in social computing, it would provide more possibilities for social data analysis, relationship discovery, outlier detection, and prediction, and is proving to be an emerging and promising direction for AI and big data research.

Measuring the space area of obstacles is one of the important problems in obstacle localizing fields. Most of the existing research works on the localization of obstacles focus on where the obstacles are, and few of them measure both the positions and the areas of the obstacles. In this paper, we propose a Minimum convex bounding Polygon localizing algorithm based on Visible light Tracking (MPVT) in order to rapidly and accurately locate the position and area of a 2D obstacle in the environment of sparsely-deployed sensors. MPVT first determines the initial localization light by Visible Light Tracing method (VLT). Second, it searches for the first side of the Minimum Convex Bounding Polygon (MCBP) of the obstacle. Third, MPVT calculates the subsequent other sides and the vertexes of MCBP until the next side coincides with the first side. In order to evaluate the approximation degree between the actual values and the localization values in terms of areas, positions and shapes, we propose two performance evaluation indexes, i.e., the area ratio and the ratio of equivalent radius. We conducted experiments on the influence of obstacle orientation and sparseness of sensor deployment, the accuracy comparison with the existing methods, and the time complexity. Experiment results show that MPVT can accurately locate the position and area of the obstacle in the environment of sparsely-deployed sensors with low time overhead, and is suitable for low-cost obstacle localization applications.

The driver's driving experience is one of the important factors affecting his or her behaviors. Prior studies have noted the effect of driving experiences on drivers' eye-movements when right-turning. However, related studies usually focused on the accident scenarios, for drivers' eye-movements on daily driving when facing right-turn, the effects of drivers' experience remain unknown. Therefore, according to design and apply a set of experiments, this paper focused on the analysis of driver's eye-movements during the daily right-turn task to compare the differences between Experienced and Novice drivers. A total of 10 drivers were invited and be classified into two groups (Experienced and Novice) to participate in the experiments. All of the drivers are driving on the right-hand side of the road, and the steering wheel is on the left side of the vehicle. The aimed data were collected by a set of glasses type eye-tracker and be further classified into four driving vision-based AOI (Areas of Interest). The results of Mann-Whitney U-tests showed that Novice drivers have a disordered line of sight and tend to spend more attention on their right view and switch their line of sight back and forth between the AOIs. Moreover, Experienced drivers more tend to keep their view directly in front of their heads instead of using the peripheral vision. The results of this study may provide guidelines to prevent accidents in Advanced Driver Assistance Systems (ADAS) and offers useful insights for the training of new drivers.

<title>Abstract</title>A color is a powerful tool used to attract people’s attention and to entice them to purchase a product. However, the way in which a specific color influences people’s color selection and the role of their eye movements and cultural factors in this process remain unknown. In this study, to delve into this problem, we designed an experiment to determine the influence of specific colors on people’s product preferences by using an eye-tracking device, intending to identify the role of their eye movements and cultural factors. Based on the experimental data, a detailed influence path model was built to describe the effect of specific colors on product evaluations by an integrated moderation and mediation analysis. Our findings show that in the influence process, the effects of specific colors on product evaluations are mediated by eye movements. Additionally, cultural factors partly moderate the process as an influencing factor. The research findings from this study have important implications for user-centered product design and visual marketing management.

Social platforms are full of rumors (i.e., unverified contents). Naturally, it is imperative but challenging to effectively determine the veracity of these rumors on popular social platforms. Previously deep learning based rumor verification schemes usually treat the issue as an independent and single task. Considering the rumor verification and stance classification are relevant tasks, we propose an effective Rumor verification scheme based on Multi-task learning Model, RV-ML, in which the shared long-short term memory (LSTM) layer for both rumor verification and stance classification can effectively deal with the sequential information for the original input, and generate macro-level virtual features, and the convolution neural network (CNN) layer uniquely designed for rumor verification task is used to mine local features from shared LSTM layer. Comparisons between our RV-ML and several typical rumor verification schemes on the real RumourEval and PHEME datasets demonstrate that our proposed scheme gains better performance for the task of rumor verification.

Many accidents are caused by drivers abruptly braking, or by their braking-related behaviors. However, it is challenging to collect data on the pre-braking behaviors of drivers by using sensors installed on vehicles, because of which this issue has not been adequately studied. In this paper, we focus on the differences in pre-braking behaviors between experienced and novice drivers by analyzing data collected from a motion capture device attached to them. The results of experiments and the Mann-Whitney U-test show that experienced drivers usually set their feet farther than novice drivers from the braking pad to reduce the chances of pressing it by mistake. However, when they needed to brake, their right foot had a lower response time for pushing the braking pad than novice drivers. The results of this study can provide guidelines for preventing accidents in the development of advanced driver assistance systems and can offer useful insights for training new drivers.

With the eruption of big data, practical recommendation schemes are now very important in various fields, including e-commerce, social networks, and a number of web-based services. Nowadays, there exist many personalized movie recommendation schemes utilizing publicly available movie datasets (e.g., MovieLens and Netflix), and returning improved performance metrics (e.g., Root-Mean-Square Error (RMSE)). However, two fundamental issues faced by movie recommendation systems are still neglected: first, scalability, and second, practical usage feedback and verification based on real implementation. In particular, Collaborative Filtering (CF) is one of the major prevailing techniques for implementing recommendation systems. However, traditional CF schemes suffer from a time complexity problem, which makes them bad candidates for real-world recommendation systems. In this paper, we address these two issues. Firstly, a simple but high-efficient recommendation algorithm is proposed, which exploits users' profile attributes to partition them into several clusters. For each cluster, a virtual opinion leader is conceived to represent the whole cluster, such that the dimension of the original useritem matrix can be significantly reduced, then a Weighted Slope One-VU method is designed and applied to the virtual opinion leader-item matrix to obtain the recommendation results. Compared to traditional clusteringbased CF recommendation schemes, our method can significantly reduce the time complexity, while achieving comparable recommendation performance. Furthermore, we have constructed a real personalized web-based movie recommendation system, MovieWatch, opened it to the public, collected user feedback on recommendations, and evaluated the feasibility and accuracy of our system based on this real-world data.

In the study of influence maximization in social networks, the speed of information dissemination decreases with increasing time and distance. The investigation of the characteristics of information dissemination is of great significance to the management and control of public opinion. A three-hop velocity decay propagation model is proposed to determine the propagation speed in information dissemination and the time and distance attenuation factors of information dissemination were modeled. We simulated the three-hop information propagation and developed an influence maximization algorithm based on the rate attenuation propagation model (IMMRA). Experiments using two example data sets showed that the proposed algorithm had higher accuracy and time efficiency than a greedy algorithm.

Using Computer-aided Diagnostic (CAD) to analyze medical images is currently a focused area, and deep learning is widely used in the detection of pulmonary nodules in medical imaging. Current detection algorithms are effective in detecting large pulmonary nodules, but their detection effect on small nodules and micro-nodules is not ideal. In order to solve this problem, this paper uses high-resolution network (HRNet) as the backbone network of Cascade R-CNN to improve its detection accuracy on small targets. HRNet can preserve the information of small target nodules in the feature map with high resolution and obtain a finegrained feature map for the detection task. This paper also combines dilated convolution with HRNet and proposes an improved HRNet named dilated HRNet. Experiments on the LIDC-IDRI dataset show that the improved Cascade R-CNN increases the detection accuracy of pulmonary nodules, especially on small nodules.

As lung cancer continues to threaten human health, Computer-Aided Diagnostic (CAD) plays an increasingly significant role in lung cancer diagnosis, and convolutional neural networks (CNNs) have shown the outstanding performance in image segmentation. In this work, Hybrid Task Cascade (HTC) is used to segment lung nodules that are difficult to find in CT images. Considering that lung nodules are usually quite small, this study integrates Feature Pyramid Network (FPN) into ResNet-50 to make full use of multi-scale feature and improve the segmentation accuracy of small target nodules. In addition, given that existing defects in Region Proposal Network (RPN), which refers to most of generated anchors are irrelevant to target objects, and the conventional method are unaware of the shapes of target objects, this work proposes to use Guided Anchoring to replace RPN in HTC and generate anchors more effectively. Experimental results on the LIDC-IDRI dataset demonstrate that the modified HTC improves the segmentation accuracy of lung nodules.

Hyper-parameters in deep learning are sensitive to prediction results. Non-maximum suppression (NMS) is an indispensable method for the object detection pipelines. NMS uses a pre-defined threshold algorithm to suppress the bounding boxes while their overlaps are not significant. We found that the pre-defined threshold is a hyper-parameter determined by empirical knowledge. We propose an adaptive threshold NMS that uses different thresholds to suppress the bounding boxes whose overlaps are not significant. The proposed adaptive threshold NMS algorithm provides improvements on Faster R-CNN with the AP metric on pneumonia dataset. Furthermore, we intend to propose more methods to optimize the hyper-parameters.

The prebraking-related actions typically studied are the main maneuvers carried out to avoid collision. Especially for those braking actions taken when turning or parking, accidents often occur because of human errors such as the incorrect choice of pedal. However, regarding these daily braking-related driving behaviors, the effects of the driver characteristics, such as driving experience and gender, on the prebraking behaviors remain unknown. Therefore, defining prebraking behaviors as the movements of a driver's body before his or her foot touches the brake pedal, this paper identifies the details of drivers' driving behaviors while prebraking by analyzing the data collected from a wearable high-precision 23-joint motion capture device and further confirms the effects of driver experience, gender and stature on these behaviors. According to two-way analyses of variance (ANOVAs) that were performed on 100 sets of motion data collected from a set of driving experiments involving two different tasks, drivers perform similar prebraking body actions even under different braking scenarios. Moreover, the results of an interaction effects analysis confirmed the impact of drivers' experiences, gender and stature on their prebraking actions. The results of this study can serve as guidelines for future self-driving and advanced driver assistance system (ADAS) development and provide useful insights for the identification and training of new drivers.

Underlying latent factors may cause a person to feel unwell. As the influence of the latent factors increases, the person will become sick. It is difficult to directly measure the influence of latent factors on risk degrees. However, early symptoms of a disease may affect vital signs such as body temperature and blood pressure, which may be a result of the influence of the latent factors. Deep learning is often used to predict the onset of a disease owing to its high accuracy. However, the reliability of this method is limited because of its characteristics of a black-box model. In this study, we propose a new approach to detect health abnormality. We regard the degree of influence of latent factors as the risk of disease and detect health abnormality before the onset of the disease. In our approach, we used a combination of Structural Equation Modeling (SEM) and Hidden Markov Model (HMM). First, in SEM, a domain model was created, and the factor score was estimated based on the relationship between latent factors and the explicit variables influenced by the factors. Thereafter, risk degrees were quantified with HMM using the estimated factor scores, and abnormality was identified in terms of risk degree. Finally, our proposed method was compared with three baselines: PCA (principal component analysis)-based approach, deep learning, and no-degree estimation methods. The average recall of our method was 98.75%, almost the same as the baselines, and false positive rate (FPR) was 0.186%, lower than the baselines. In the five-fold cross-validation comparing with no-degree estimation method, the average accuracy and recall of our method were 99.7% and 98.3% respectively, and FPR was 0.045%, all much better than the baseline. Moreover, our approach can make the process of obtaining the result visible and help detect abnormality sensitively by setting the threshold according to the risk degree, which can contribute to early detection of a disease and improve the reliability of abnormality detection as well.

Background: Cost-sensitive algorithm is an effective strategy to solve imbalanced classification problem. However, the misclassification costs are usually determined empirically based on user expertise, which leads to unstable performance of cost-sensitive classification. Therefore, an efficient and accurate method is needed to calculate the optimal cost weights. Results: In this paper, two approaches are proposed to search for the optimal cost weights, targeting at the highest weighted classification accuracy (WCA). One is the optimal cost weights grid searching and the other is the function fitting. Comparisons are made between these between the two algorithms above. In experiments, we classify imbalanced gene expression data using extreme learning machine to test the cost weights obtained by the two approaches. Conclusions: Comprehensive experimental results show that the function fitting method is generally more efficient, which can well find the optimal cost weights with acceptable WCA.

(accepted)

Today, thanks to the development of wearable devices, people can easily accumulate and get access to their own health data. Through these health data, individuals can better quantify and manage their health status. However, there is a problem that health data is not fully used due to its complexity that non-expert individuals experiencing difficult in understanding. Visualization is widely used in helping people to understand data. In this study, we design a visualization system to solve the problem. Based on the intuitive design and intelligible interaction concept, we propose four functions: overview, detailed annotation, temporal comparison and horizontal comparison in the visualization system. We utilize persona analysis to help interaction design in the visualization system and describe application scenarios.

With more and more people suffer from lung cancer, computer-aided diagnosis plays a more and more important role in lung cancer diagnosis. CNN has achieved state-of-the-art performance in image processing, and Mask R-CNN outperforms most other methods on instance segmentation. However, the target is extraordinarily small, and the background is very large in images, which results in a large number of negative examples and most of them are easy negatives. They will contribute a large part of the loss value in smooth loss function. The class imbalance problem leads to inefficient training, which makes model degenerated. In this paper, we propose a method based on Mask R-CNN to segment lung nodules. Due to the non-uniformity of CT values, we use the Laplacian operator to do feature dimensionality reduction for filtering out part of the noise. In our model, the novel function Focal Loss is used to suppress well-classified examples. The model is tested on LIDC-IDRI dataset and the results showed that the average precision of lung nodules reaches 78%. Compared with the smooth loss function in Mask R-CNN it improves by 7%.

With the development of eye-tracking technology, existing studies verified the effect of culture on people' s color preference by their eye-movements. However, prior studies usually see Asians to be holistic, and for specific colors, the details of these effect's type and influence paths are remains in black box. In this paper, we focus on the difference between Chinese and Japanese who have similar but different cultures and try to identify the specific colors' influence paths which have affect their final product selecting by analyzing the data collected from eye-tracking devices. Base on the results of mediation and moderation tests, the finding proves that people's eye-movement metrics have mediated the effects of specific colors on their products selection, and their culture as an important moderation factor have moderate the same process partly. The results can provide guidelines for products design or visual marketing management.

Twitter, as a popular social networking service, is used all over the world, with which users post tweets for various purposes. When users post tweets, an emotion may be behind the messages. As the emotion changes over time, we should better consider their emotional changes and states when analyzing the tweets. In this study, we improve polarity classification by considering the poster’s emotional state. Firstly, we analyze the sentence structure of a tweet and calculate emotion scores for each category by Naive Bayes. Then, the poster’s emotion state is estimated by the emotion scores, and a prediction model of emotional state is created by Long Short Term Memory (LSTM). Based on the predicted emotional state, weights are added to the scores. Finally, polarity classification is performed based on the weighted emotion scores for each category. In our experiments, our approach showed better accuracy than other related studies.

Recently, location-based social networks (LBSNs) have witnessed a significant development, and have attracted millions of mobile users to share their locations and location-related contents. Due to the huge data in LBSNs, it is imperative to design an efficient location recommendation scheme for abundant users. But most of existing works suffer from poor accuracy and efficiency. On one hand, the traditional user-based collaborative filtering (CF) methods only focus on user characteristics, which limit the recommendation accuracy. On the other hand, it is inefficient to traverse all LBSN data in each recommendation. To solve the above issues, this paper proposes a new location recommendation method LC–G–P, in which, Louvain method is adopted to cluster users into several communities; and meanwhile, multiple features including Geographical distance, location Popularity are incorporated into location recommendation. Experiments based on real LBSN dataset illustrate that the proposed recommendation scheme LC–G–P has better accuracy and efficiency, in comparison with the existed typical location recommendation schemes.

Nowadays, along with the high development of emerging computational paradigms, more and more populations have been involved into the social revolution across various intelligent systems, which results in dynamic user connections associated with a variety of social behaviors. The associated users with different properties, who can be regarded as one kind of information resources, have become increasingly important, especially in social knowledge creation and human intelligence utilization processes. In this study, we concentrate on user role identification based on their social connections and influential behaviors, in order to facilitate information sharing and propagation in social networking environments. Following the construction of a dynamic user networking model, we propose a network-aware method to identify four kinds of special users, who may play an important role in information delivery among a group of users, or knowledge sharing between pairs of users. A set of attributes and measures is proposed and calculated to identify and represent these users based on the analysis of their influence-related social behaviors and dynamic connections. Experiments and evaluations are conducted to demonstrate the practicability and usefulness of the proposed method using Twitter data. Analysis results show the effectiveness of our approach in identifying the distinct features of four kinds of users from the user networking model. Comparison experiments indicate that the proposed identification method outperforms two other related works. Finally, a questionnaire-based evaluation demonstrates the accuracy and efficiency of the proposed method in terms of finding these users in a real social networking environment.

For the benefit from accurate electricity price forecasting, not only can various electricity market stakeholders make proper decisions to gain profit in a competitive environment, but also power system stability can be improved. Nevertheless, because of the high volatility and uncertainty, it is an essential challenge to accurately forecast the electricity price. Considering that recurrent neural networks (RNNs) are suitable for processing time series data, in this paper, we propose a bidirectional long short-term memory (LSTM)-based forecasting model, BRIM, which splits the state neurons of a regular RNN into two parts: the forward states (using the historical electricity price information) are designed for processing the data in positive time direction and backward states (using the future price information available at inter-connected markets) for the data in negative time direction. Moreover, due to the fact that inter-connected power exchange markets show a common trend for other neighboring markets and can provide signaling information for each other, it is sensible to incorporate and exploit the impact of the neighboring markets on forecasting accuracy of electricity price. Specifically, future electricity prices of the interconnected market are utilized both as input features for forward LSTM and backward LSTM. By testing on day-ahead electricity prices in the European Power Exchange (EPEX), the experimental results show the superiority of the proposed method BRIM in enhancing predictive accuracy in comparison with the various benchmarks, and moreover Diebold-Mariano (DM) shows that the forecast accuracy of BRIM is not equal to other forecasting models, and thus indirectly demonstrates that BRIM statistically significantly outperforms other schemes.

With the help of fog computing, urban computing and intelligence novel systems can be created to improve the urban environment and the quality of human life. Sensor-cloud systems based on urban fog computing (SCS-UFC) are new intelligent network systems, which combine a cloud platform with wireless sensor networks (WSNs) as well as fog nodes to provide functions such as sensing, computation, and storage of large-scale data. Since the sensor nodes in WSNs only have limited transmission capacity, they cannot transmit their data to the cloud platform directly. Therefore, fog nodes with stronger transmission capacity are deployed to relay the data from WSNs to the cloud platform. However, different fog nodes may be burdened with different workloads (i.e., amounts of data): usually, the fog nodes with heavier workloads mean longer transmission delay and more energy consumption. If a fog node exhausts its energy, it will die and then make the network cease to work. Therefore, it is necessary to balance the workload of all fog nodes so as to reduce transmission delay and energy consumption of the sensors. However, addressing the problem is challenging because each fog node only knows local information of its neighbors, and thus it is difficult to get a global optimization result by itself. In this paper, a distributed intelligent algorithm based on the Hungarian method is proposed. First, each fog node collects the information connected with its neighboring fog nodes that are located within its transmission range. Then, a new genetic algorithm is designed to find an approximate optimization solution. Finally, each fog node decides if it should forward parts of its workload to other fog nodes so that the workloads of all fog nodes are balanced. Simulation results show that our algorithm can achieve shorter delay and less energy consumption than existing works.

Smart grids are vulnerable to cyber-attacks, which can cause significant damage and huge economic losses. Generally, state estimation (SE) is used to observe the operation of the grid. State estimation of the grid is vulnerable to false data injection attack (FDIA), so diagnosing this type of malicious attack has a major impact on ensuring reliable operation of the power system. In this paper, we present an effective FDIA detection method based on residual recurrent neural network (R2N2) prediction model and adaptive judgment threshold. Specifically, considering the data contains both linear and nonlinear components, the R2N2 model divides the prediction process into two parts: The first part uses the linear model to fit the state data; the second part predicts the nonlinearity of the residuals of the linear prediction model. The adaptive judgment threshold is inferred through fitting the Weibull distribution with the sum of squared errors between the predicted values and observed values. The thorough simulation results demonstrate that our scheme performs better than other prediction based FDIA detection schemes.

With the development of eye-tracking technology, existing studies verified the effect of culture on eye movements. However, the detail of how the different cultures, especially the different Asian cultures affect people’s color preferences about products by visual attention remains in black box. In this paper, we focus on the difference between Chinese and Japanese who have similar but different cultures and try to identify their difference in visual attention about color preferences when selecting goods by using eye-tracking. The finding proves that people with different Asians cultures have different viewing patterns when faced to select goods with different colors. The results also indicate that culture as a factor does affect to ones’ evaluation of products. The results can provide guidelines for products design or visual marketing management.

© 2019 Elsevier GmbH Introduction: The utilization of information and communications technology (ICT) devices is a new way to record and analyze health data of the elderly. This time-series study aimed to analyze health changes, and the correlation between pulse manifestation and health indicators, in the elderly. Methods: We conducted continuous 93-day monitoring of health data and pulse records in 8 elderly participants. A time series method was used to analyze health changes in participants. The correlation between pulse manifestation and health indicators was analyzed using a structural equation model. Results: During the use of ICT devices, the number of steps taken by the elderly showed a significant increase (p < 0.05). According to the time series prediction formula, the number of steps predicted on day 94 was 8869.8. In practice, steps on day 94 were 8267.3; the difference between these values was within 10%. Personal health information, health habits, and physiological indicators had a direct impact on pulse manifestation; influence coefficients were 0.14, 0.18, and 0.05 respectively. Conclusion: Through the use of ICT devices, we can understand the health status of the elderly and make behavior predictions. Pulse manifestation data can indicate the health status of the elderly. Thus, ICT devices can be used as health management tools and assist doctors in making simple diagnoses.

With the emergence and wide application of cyber technologies, the process of medical informatization has progressed rapidly in recent years. The collection of gene expression data and cyber-enabled tumor risk analysis has matured and is becoming more common. In the case of tumor risk analysis, identification of the distinct genes that contribute the most to the occurrence of tumors has become an increasingly important issue. In this paper, based on gene selection, an improved SSO (Simplified Swarm Optimization) algorithm is developed for data-driven tumor risk analysis that is able to obtain a higher classification accuracy with fewer selected genes. The proposed algorithm is called iSSO-HF&LSS (improved SSO with a hybrid filter and local search strategy) and utilizes information gain and the Pearson correlation coefficient as a hybrid filter method to select a small number of distinct and discriminative genes. Moreover, to select an optimal gene subset, a new local search strategy is applied. The proposed local search strategy selects informative but fewer correlated genes by considering their correlation information. To evaluate the efficiency of the algorithm, a series of experiments is conducted using ten tumor gene expression datasets, and a comparison is made between the performance of this proposed method and nine well-known benchmark classification methods as well as methods used in six referenced studies. As evaluated by several statistical analyses, the proposed method outperforms the existing methods with significant differences and efficiently simplifies the number of gene expression levels.

Conventional online learning typically allows an instructor to deliver instruction to students via a predefined curriculum and within a fixed knowledge structure (i.e., explaining the instructional subject). With the dramatic growth of social media technology and correlated data aggregation, some sort of instant knowledge is obtained by daily users. An emerging type of knowledge (i.e., social knowledge) has been identified and may lead to self-paced learning from social networks, which is simply defined as social learning. This article points out three important issues for social learning, namely, knowledge retrieval via temporal social factors, and the connection between social network and the knowledge domain. Two significant automation mechanisms, lecture generation for self-regulated learning and influencing domain computation for opportunity finding, are suggested to facilitate the process of social learning. A prototype system based on Elgg was implemented, sourced by a federated repository that has stored and shared more than 1.5 millions transactions (e.g., content, interactions, etc.). We conclude that timely social knowledge (or crowdsourcing results) can be widely applied in the next era of online learning environment. Findings through the statistical analysis are prospective to support understanding of phenomenon of social learning and design of future learning platform for followup researchers.

Process mining is a technology to gain knowledge of the business process by using the event logs and achieve a model of the process, which contributes to the detection and improvement of the business process. However, most existing process mining algorithms have drawbacks associated with managing uncertain data, and the method of using the frequency threshold alone needs to be enhanced. This paper improves correlation measures in heuristic mining to build a correlation matrix based on an improved frequency matrix. Combined with the maximum entropy principle, a self-adaptive method to determine the threshold is given, which is used to remove the uncertain data relationship in the logs. Furthermore, this study identifies a selective and parallel structure through a modified frequency matrix, and we can get a Petri net-based process model from a directed graph. The recognition of parallel structures contributes to eliminating imbalances when calculating the threshold to deal with the uncertain data. Finally, this paper presents an algorithm framework for adaptively removing uncertain data. This study represents a new attempt to use entropy to remove uncertain data in the field of Business Process Management (BPM). The threshold to deal with the uncertain data does not need to set the parameters in advance. Therefore, the proposed algorithm is self-adaptive and universal. Experimental results show that the algorithm proposed in this study has a higher degree of behavioral and structural appropriateness, and fitness, for the uncertain log data compared to traditional algorithms.

In recent years, deep learning has made great progress in image classification and detection. Popular deep learning algorithms rely on deep networks and multiple rounds of back-propagations. In this paper, we propose two approaches to accelerate deep networks. One is expanding the width of every layer. We reference to the Extreme Learning Machine, setting big number of convolution kernels to extract features in parallel. It can obtain multiscale features and improve network efficiency. The other is freezing part of layers. It can reduce back-propagations and speed up the training procedure. From the above, it is a random convolution architecture that network is proposed for image classification. In our architecture, every combination of random convolutions extracts distinct features. Apparently, we need a lot of experiments to choose the best combination. However, centralized computing may limit the number of combinations. Therefore, a decentralized architecture is used to enable the use of multiple combinations.

The rapid development of the Internet of Things (IoT) and e-commerce has brought a lot of convenience to people's lives. IoT applications generate a large number of services and user data. It is necessary to design a personalized recommendation technology suitable for the users of IoT services and improve the user experience. In this paper, a recommendation algorithm with trusted relevance combined with matrix factorization is proposed. By establishing an effective trust metric model, the user's social information is integrated into the recommendation algorithm. First, the social network concentric hierarchical model is used to consider the direct or indirect trust relationship, and more trust information is integrated for the matrix factorization recommendation algorithm. Then, we design the trust relevance, comprehensively considering the trust factors and interest similar factors. Our experiments were performed on the Dianping datasets. The recommendation algorithm using matrix factorization and trusted relevance degree has higher prediction accuracy than the basic matrix decomposition and social matrix factorization in terms of accuracy and stability.

Fast development of shared services has become a crucial part of the cyber-enabled world construction process, as sharing services reinvent how people exchange and obtain goods or services. However, privacy leakage or disclosure remains a key concern during the sharing service development process. While significant efforts have been undertaken to address various privacy issues in recent years, there is a surprising lack of review for privacy concerns in the cyber-enabled sharing world. To bridge the gap, in this paper, we survey and evaluate existing and emerging privacy issues relating to sharing services from various perspectives. Differing from existing similar works on surveying sharing practices in various fields, our work comprehensively covers six branches of sharing services in the cyber-enabled world and selects solutions mostly from the recent five to six years. We conclude the issues and solutions from three perspectives, namely, from users', platforms' and service providers' perspectives. Hot topics and less discussed (cold) topics are identified, which provides hints to researchers for their future studies.

Feature selection (FS) is one of fundamental data processing techniques in various machine learning algorithms, especially for classification of healthcare data. However, it is a challenging issue due to the large search space. This paper proposed a confidence based and cost effective feature selection method using binary particle swarm optimization, CCFS. First, CCFS improves search effectiveness by developing a new updating mechanism, in which confidence of each feature is explicitly considered, including the correlation between feature and categories, and historically selected frequency of each feature. Second, the classification accuracy, the feature reduction ratio, and the feature cost are comprehensively incorporated into the design of the fitness function. The proposed method has been verified in UCI cancer classification dataset (Lung Cancer). The experimental result shows the effectiveness of the proposed method, in terms of accuracy and feature selection cost.

© 2018 IEEE. This study used wearable devices and traditional Chinese medicine (TCM) diagnosis, which verifies the promotion of wearable devices on the health habits of the elderly, and the correlation between pulse diagnosis and health. We used wearable devices and TCM diagnosis (including TCM doctor diagnosis and Pulse diagnosis instrument) to record health information for the elderly. One-way repeated measures analysis of variance (ANOVA) was used to determine health changes. Correlation analysis was used to determine the relationship between pulse and health indicators. There was significant change in the number of steps for the elderly before and after using the wearable device. Systolic blood pressure (SBP) was negatively correlated with pulse wave time. Diastolic blood pressure (DBP) was negatively correlated with pulse duration. Pulse diagnosis can be used as an indicator to report the health of the elderly. And used wearable devices can quantify health indicators and improve the health awareness.

Recently, many universities, particularly in China, started to offer undergraduate degree programs for the discipline of 'Intelligence Science and Technology'. However, until now, there is no universally recognized, explicitly defined, standard curriculum for the discipline of 'Intelligence Science and Technology'. This is probably because it is very difficult for scientists from various disciplines (including Computer Science, Artificial Intelligence, Cognitive Science, etc.) to reach a universally accepted answer to the fundamental question, what is intelligence and why study it? Based on our understanding and consideration about the discipline of 'Intelligence Science and Technology', in this paper we sketch a primary framework for the body of knowledge about the discipline of 'Intelligence Science and Technology', propose a curriculum model for undergraduate degree programs of the discipline, and show example curriculums. Many practical curriculums for the discipline of 'Intelligence Science and Technology' can be derived and developed from our curriculum model according to different emphases on a specific scientific or technical aspect in different departments.

Collaboration is increasingly becoming a vital factor in an academic network, which can bring lots of benefits for scholars. Ubiquitous intelligence also provides an effective way for scholars to find collaborators. However, due to the large-scale of scholarly big data, there is a lot of information hard to capture in networks and we need to dig out valid information from collaboration networks. It is a valuable and urgent task to find appropriate collaborators for scholars. To address these problems, we hypothesize that fusing topic model and structure information could improve the performance of recommendations. In this paper, we propose a collaborator recommendation system, named TNERec (Topic-aware Network Embedding for scientific collaborator Recommendation), learning representations from scholars' research interests and network structure. TNERec first extracts scholars' research interests based on topic model and then learns vectors of scholars with network embedding. Finally, top-k recommendation list is generated based on the scholar vectors. Experimental results on a real-world dataset show the effectiveness of the proposed framework compared with state-of-the-art collaboration recommendation baselines.

Since visual design of Graphical User Interface (GUI) is one of the indispensable parts of every software and to some extent, it determines whether the software is attractive. And with the increasing amount of GUI designs, it is usually impractical to deal with a variety of designs manually with limited staff. Therefore, a solution for that program is needed. This work designs and implements an automatic GUI testing system based on Deep Neural Network (DNN) and cloud service. The system includes screenshot capturing and uploading module, web server, DNN models, etc. It provides accurate and efficient GUI automatic testing technology.

Support Vector Machine (SVM) is a set of machine learning algorithms, which has been widely used in diverse domains. With the increasing size of datasets, the traditional SVM training algorithms for large-scale datasets become infeasible. Mathematical optimization and cascade parallelism are both popular strategies for accelerating SVM training. In these parallel methods, the use of appropriate parallel framework to reduce SVM training time has become a priority issue and a research focus. In this paper, we investigate and overview mathematical optimization algorithms and parallel technologies of SVM, and summarize parallel SVM solutions and application problems under different frameworks. We propose a Vote Parallel SVM to reduce the training time. Finally, we show experimental results comparing with baseline methods.

Wi-Fi fingerprint-based indoor localization is regarded as one of the most promising techniques for location-based services. However, it faces serious problem of privacy disclosure of both clients' location data and provider's fingerprint database. To address this issue, this letter proposes a differential privacy (DP)-based privacy-preserving indoor localization scheme, called DP3, which is composed of four phases: access point (AP) fuzzification and location retrieval in client side and DP-based finger clustering and finger permutation in server side. Specifically, in AP fuzzification, instead of providing the measured full finger (including AP sequence and the corresponding received signal strength), a to-be-localized (TBL) client only uploads the AP sequence to the server. Then, the localization server utilizes the DP-enabled clustering to build the fingerprints related to the AP sequence into $k$ clusters, permutes these reference points in each cluster with exponential mechanism to mask the real positions of these fingerprints, and sends the modified data set to the TBL client. At client side, location retrieval phase estimates the location of the client. Theoretical and experimental results show that DP3 can simultaneously protect the location privacy of the TBL client and the data privacy of the localization server.

The promise of cross-disciplinary scientific collaboration has recently been proven by both technological innovation and scientific research. Much effort has been spent on research collaboration recommendation. A remaining challenge is to make valuable recommendation to specific researchers in specific fields in order to obtain more fruitful cross-disciplinary collaboration. Cross-disciplinary information hides in big data and the relationships between different fields are complicated, complex, and subtle. This paper proposes a method for cross-disciplinary collaboration recommendation (CDCR) to analyze cross-disciplinary collaboration patterns in scholarly big data, and recommend valuable research fields for possible cross-disciplinary collaboration. A cross-disciplinary discovery algorithm based on topic modeling is designed to extract potential research fields. Collaboration patterns are examined by analyzing the research field correlations. A recommendation algorithm is developed to provide a specific recommendation list of potential research fields according to the discovered cross-disciplinary collaboration patterns with researchers’ profiles. Evaluations conducted based on a real scholarly dataset demonstrate the effectiveness of the proposed method in recommending potentially valuable collaborations.

Many human-centric systems have begun to use business process management technology in production. With the operation of business process management systems, more and more business process logs and human-centric data have been accumulated. However, the effective utilization and analysis of these event logs are challenges that people need to solve urgently. Process mining technology is a branch of business process management technology. It can extract process knowledge from event logs and build process models, which helps to detect and improve business processes. The current process mining algorithms are inadequate in dealing with log noise. The family of alpha-algorithms ignores the impact of noise, which is unrealistic in real-life logs. Most of the process mining algorithms that can handle noise also lack reasonable denoising thresholds. In this paper, a new assumption on noise is given. Furthermore, an anti-noise process mining algorithm that can deal with noise is proposed. The decision rules of the selective, parallel, and non-free choice structures are also given. The proposed algorithm framework discovers the process model and transforms it into a Petri network representation. We calculate the distance between traces to build the minimum spanning tree on which clusters are generated. The traces of the non-largest clusters are treated as noise, and the largest cluster is mined. Finally, the algorithm can discover the regular routing structure and solve the problem of noise. The experimental results show the correctness of the algorithm when compared with the α ++ algorithm.

People' visual information when walking on stairs is regarded an important variable that may affect their posture and balance in previous studies, but the detail of their eye movement remains unclear. In this study, we focus on people's eye-movement when walking on stairs and design an experiment to examine subjects' visual information (visual field and fixation) and possible influencing factors by using a mobile eye tracker. Based on data analysis, our results proved that the movement of going up/down stairs are significantly related to subjects' visual field. The results also indicated that the factor of subjects' gender affects both visual field and average number of fixation, but the factor of load-carrying situation only affects visual field. Some foundational insights of this study can be used to detect the fall risk for the elderly.

With smartphones increasingly becoming ubiquitous and being equipped with various sensors, nowadays, there is a trend towards implementing HAR (Human Activity Recognition) algorithms and applications on smartphones, including health monitoring, self-managing system and fitness tracking. However, one of the main issues of the existing HAR schemes is that the classification accuracy is relatively low, and in order to improve the accuracy, high computation overhead is needed. In this paper, an efficient Group-based Context-aware classification method for human activity recognition on smartphones, GCHAR is proposed, which exploits hierarchical group-based scheme to improve the classification efficiency, and reduces the classification error through context awareness rather than the intensive computation. Specifically, GCHAR designs the two-level hierarchical classification structure, i.e., inter-group and inner-group, and utilizes the previous state and transition logic (so-called context awareness) to detect the transitions among activity groups. In comparison with other popular classifiers such as RandomTree, Bagging, J48, BayesNet, KNN and Decision Table, thorough experiments on the realistic dataset (UCI HAR repository) demonstrate that GCHAR achieves the best classification accuracy, reaching 94.1636%, and time consumption in training stage of GCHAR is four times shorter than the simple Decision Table and is decreased by 72.21% in classification stage in comparison with BayesNet.

The growth of cyberspace brings more information to service recommendation. The scores of item are used in most of the recommendation algorithms, but the attributes of users and items are rarely involved in trust recommendation in cyberspace. Both the rating features and attribute information are important for trust recommendation results. In this paper, we combine the heterogeneous information in cyberspace and propose a novel trust recommendation model based on the latent factor model and trusty neighborhood fitting model. We utilize the feature based Latent Factor Model and study the linear features integrated model To solve the failure problem of the latent factor model in the integrated model under the cold-start situations, we propose two optimized methods, which contain the filling method based on feature similarity and the filling method based on feature regression through mapping attributes to features. Experimental results show that the improved method outperforms traditional collaborative recommendation in terms of recommendation accuracy. Meanwhile, our proposed method has been verified to free from the impact of the cold-start problem.

The rapid development of the Internet and ecommerce has brought a lot of convenience to people's life. Personalized recommendation technology provides users with services that they may be interested according to users' information such as personal characteristics and historical behaviors. The research of personalized recommendation has been a hot point of data mining and social networks. In this paper, we focus on resolving the problem of data sparsity based on users' rating data and social network information, introduce a set of new measures for social trust and propose a novel personalized recommendation algorithm based on matrix factorization combining trust relevancy. Our experiments were performed on the Dianping datasets. The results show that our algorithm outperforms traditional approaches in terms of accuracy and stability.

Considering that the challenges in using peer grading to select the best-k students in MOOCs (massive open online courses) are twofold: first is strategyproof, i.e., students should not benefit by untruthfully reporting valuations; second, instead of exerting (costly) efforts to evaluate, students may randomly provide (or just guess) the evaluations on other peers. This paper proposes a truthful grading-based strategyproof peer selection scheme for MOOCs, TSP. Specifically, all students are partitioned into same-size clusters, and each student only evaluates students in other clusters. Moreover, peer prediction mechanism was utilized to motivate each student to truthfully report their gradings through comparing their reports with other peer students who conduct same evaluation tasks. The theoretical analysis and simulation results show that TSP can both stimulate students to truthfully reporting their gradings, and meanwhile select best k-students in strategyproof way.

With the rapid development of Internet of Things (IOT) technologies, modern power systems have become complex cyber-physical systems. A large number of smart devices have promoted efficient generation, transmission and distribution in the smart grid. State estimation (SE) is one of fundamental components in smart grid that evaluates the operation state of a grid by using a set of sensor measurements and grid topologies. A major issue is the authenticity of the measurements collected by the sensors. Specifically, the false data injection attack (FDIA) aims to temper the information that reflect the grid operation state. In this paper, we propose an efficient prediction-based FDIA detection and location scheme, PDL, in which the state vector of smart grid can be represented as multivariate time series, and can be predicted by vector autoregressive processes (VAR) through intentionally exploiting the temporal and spatial correlations of states. Different from most previous works which assumed the state transfer matrix constant and diagonal, a time-varying and non-diagonal matrix is adopted in this scheme. Then, the consistency between the predicted measurements and the observed measurements is utilized to detect and locate abnormal data. Besides, the detected abnormal data can be replaced with the predicted data, which simplifies the calibration process. Extensive simulation results verify the performance of the proposed scheme.

Owing to advanced storage and communication capabilities today, smart devices have become the basic interface between individuals and their surrounding environment. In particular, massive devices connect to one other directly in a proximity area, thereby enabling abundant Proximity Services (ProSe), which can be classified into two categories: public safety communication and social discovery. However, two challenges impede the quick development and deployment of ProSe applications. From the viewpoint of networking, no multi-hop connectivity functionality component can be directly operated on commercially off-the-shelf devices, and from the programming viewpoint, an easily reusable development framework is lacking for developers with minimal knowledge of the underlying communication technologies and connectivity. Considering these two issues, this paper makes a twofold contribution. First, a multi-hop mesh networking based on Bluetooth Low Energy (BLE) is implemented, in which a proactive routing mechanism with link-quality (i.e., received signal strength indication) assistance is designed. Second, a ProSe development framework called BLE Mesh is designed and implemented, which can provide significant benefits for application developers, framework maintenance professionals, and end users. Rich application programming interfaces can help developers to build ProSe apps easily and quickly. Dependency inversion principle and template method pattern allow modules in BLE Mesh to be loosely coupled and easy to maintain and update. Callback mechanism enables modules to work smoothly together and automation processes such as registration, node discovery, and messaging are employed to offer nearly zero-configuration for end users. Finally, based on the designed ProSe development kit, a public safety communications app called QuoteSendApp is built to distribute emergency information in close area without Internet access. The process illustrates the easy usability of BLE Mesh to develop ProSe apps.

Twitter is one of the most popular social network services (SNS) applications, in which users can casually post their messages. Given that users can easily post what they feel, Twitter is widely used as a platform to express emotions. These emotional expressions are considered to possibly influence user relationships on Twitter. In our previous study, we analyzed this influence using emotional word dictionaries. However, we could not measure the emotion scores for the words not included in the dictionaries. To solve this problem, in this study, we use the Naive Bayes and consider dependency parsing, i.e., the structure of tweets and the relationships of words. Furthermore, we introduce a set of new measures, namely total positive emotion score (TPES), total negative emotion score (TNES), and total neutral emotion score (TNtES). Based on these measures, we define a new composite index (CI) for emotion scores, which is a normalized value in the range of 0 to 1. We categorize users into positive and negative groups based on the composite index and test the difference of user relationships between these two groups with a statistical method. The result demonstrates that the relationships of positive users not only get better (i.e., the number increases) with time, but also tends to be mutual, which is consistent with the result of our previous study.

Purpose: Rapid developments in information technology have enabled wearable devices to be applied in the health field. In elderly adults, wearable devices aid in data collection and exerts a positive effect on their social capital. This study evaluated the changes in these two parameters among elderly adults using wearable devices, and analyzed the effect of these devices on their daily lives. Methods: We selected 18 elderly people using wearable devices, between February and May 2017. The data collected by the wearable devices included the number of steps taken, sleep duration, blood pressure, heart rate, respiratory rate, fatigue, and mood of the wearers. Using a questionnaire and the trajectory equifinality model, we interviewed and surveyed elderly adults in order to understand their health status and social capital. Results: The health of the participants was generally good, and most were able to achieve > 8000 steps per day (p < 0.05). Mild and moderate fatigue symptoms were noted in elderly adults for 90% of the study period (p < 0.05). The number of steps, blood pressure, and heart rate changed significantly within a month. From the commencement of using the wearable devices, a steady increase was noted in the monthly number of steps. Interviews suggested that the elderly adults perceived wearable devices as having the potential to improve health and social capital. Conclusions: By using wearable devices, the participants had a better understanding of their own health, and were willing to take health-boosting measures. The participants were also more willing to increase their social capital and expand their social network.

In trustworthy service discovery for Mobile Social Networking in Proximity (MSNP), conventional trust computation faces a big challenging issue—relatively high latency. To cope with it, trustworthiness determination strategies were proposed in our previous study, aiming at avoiding trust computation under certain conditions, so as to reduce the latency. These strategies are conceived based on the assumption that data of a user’s past experience and current profile could be used, and they are incorporated with a set of thresholds from the analysis result of these data. The settings of these thresholds directly affect service quality and user satisfaction on the MSNP service, which in turn becomes an optimization problem. In this paper, we focus on formulating this optimization problem and demonstrating the effectiveness of our proposed solution by designing a simulation experiment. In detail, we establish mathematical models and adjust parameters. We conduct simulations on MATLAB and analyze the results obtained under several different settings. We further compare our work with related works. The results show that our proposed solution is practically feasible and effective in reducing latency under certain conditions.

Nowadays, the cyber, social and physical worlds are increasingly integrating and merging. Especially, combining the strengths of humans and machines helps tackle increasing hard tasks that neither can be done alone. Following this trend, this paper designs a Quality aware Truthful Incentive mechanism for cyber–physical enabled Geographic crowdsensing called Geo-QTI. Different from existing work, Geo-QTI appropriately accommodates the utilities of various stakeholders: requesters, participants and the crowdsourcing platform, and explicitly takes the requesters’ quality requirements, and participants’ quality provision into account. Geo-QTI explicitly includes four components: requester selection, participant selection, pricing and allocation. Requester selection with feasible analysis removes the requesters whose job cannot be completed by all participants or suffers from the monopoly participant (without the participant's contribution, others cannot cover requesters’ requirement), obtains winning requesters set and determines actual payments. In participant selection phase, the platform aggregates the requested tasks (submitted by all winning requesters) in the sensed geographic area, and chooses the appropriate participants satisfying the winning requesters’ quality requirements with total cost as low as possible. Pricing phase determines the payments to winning participants. The phase of allocation assigns the specific participants to minimally cover the quality requirements of those winning requesters. Rigid theoretical analysis demonstrates Geo-QTI can achieve both requesters’ and participants’ individual rationality and truthfulness, computational efficiency and budget balance for the platform. Furthermore, the extensive simulations confirm our theoretical analysis, and illustrate that Geo-QTI can reduce requesters’ expenses greatly and ensure the fairness of allocation.

Community structure is a typical feature of complex networks in cyberspace, and community detection is considered to be crucial to understanding the topology structure, network function and social dynamics of cyberspace. However, some particular nodes may simultaneously belong to several communities in cyberspace. Though there are many algorithms to detect the overlapping communities, most of them are based on the network structure without considering the attributes of the nodes. In this paper, we focus on the convergence characteristic of network and propose an overlap community detection algorithm based on the node convergence degree, which is defined as a combination of attribute convergence degree and structure convergence degree. It combines the network topology with the attributes of the nodes and considers both local and global information of a node. An improved PageRank algorithm is used to get the importance of each node in the global network, while the information of local network is used to measure the structure convergence degree. The overlap communities are thus identified by spectral cluster based on the node convergence degree. Finally, experiment results demonstrate the effectiveness and better performance of our proposed method.

One of the main purposes for which people use Twitter is to share emotions with others. Users can easily post a message as a short text when they experience emotions such as pleasure or sadness. Such tweet serves to acquire empathy from followers, and can possibly influence others' emotions. In this study, we analyze the influence of emotional behaviors to user relationships based on Twitter data using two dictionaries of emotional words. Emotion scores are calculated via keyword matching. Moreover, we design three experiments with different settings: calculate the average emotion score of a user with random sampling, calculate the average emotion score using all emotional tweets, and calculate the average emotion score using emotional tweets, excluding users of few emotional tweets. We evaluate the influence of emotional behaviors to user relationships through the Brunner-Munzel test. The result shows that a positive user is more active than a negative user in constructing user relationships in a specific condition.

Massive Online Open Courses (MOOCs) have the potential to revolutionize higher education with their wide outreach and accessibility. One of key challenges in MOOCs is the student evaluation: The large number of students makes it infeasible for instructors or teaching assistants (TAs) to grade all assignments. Peer grading-having students assess each other-is a promising approach to tackling the problem of evaluation at scale. The user evaluations are then used directly, or aggregated into a consensus value. However, lacking an incentive scheme, users have no motive in making effort in completing the evaluations, providing inaccurate answers instead. To address the above issues, we propose and implement a peer grading scheme, RankwithTA. Specifically, considering that the quality of a student determines both her performance in the assignment and her grading ability, RankwithTA makes the grade each student received depend on both the quality of the solution they submitted, and on the quality of their review and grading work to incentivize students' correct grading, Furthermore, the ground truth is incorporated, which utilizes external calibration by having some students graded by instructors or TAs to provide a basis for accuracy. The simulation results illustrate that RankwithTA performs better than the existing schemes.

Cost-sensitive algorithms have been widely used to solve imbalanced classification problem. However, the misclassification costs are usually determined empirically, leading to uncertain performance. Hence an effective method is desired to automatically calculate the optimal cost weights. Targeting at the highest weighted classification accuracy (WCA), we propose two approaches to search for the optimal cost weights, including grid searching and function fitting. In experiments, we classify imbalanced gene expression data using extreme learning machine to test the cost weights obtained by the two approaches. Comprehensive experimental results show that the function fitting is more efficient which can well find the optimal cost weights with acceptable WCA.

© 2019, Springer Nature Switzerland AG. Pulse diagnosis is a typical diagnosis of Traditional Chinese Medicine (TCM). However, it is not clear if there is any relationship between the result of pulse diagnosis and other health related data. In this study, we investigate this and analyze pulse diagnosis data from a TCM doctor and a pulse diagnostic instrument (PDI) by Random Forest. Subjects’ vital signs and pulse diagnosis data from a TCM doctor are used as training data. We classify vital signs which have the PDI’s diagnoses labels. As a result, classification accuracies were over 60% in all cases. Our experiment results imply that better pulse diagnosis may be made with assistance of personal health data analysis.

Nowadays, it has become convenient to record data related to an individual using a wearable device. However, it is difficult to utilize the data according to the individual, especially to anomaly detection. Anomaly detection is very important for healthcare, e.g., early detecting of illness. In our previous study, we proposed an approach to specifying latent factors using Structural Equation Modeling (SEM). In this paper, we propose an improved approach for anomaly detection taking account of personal status based on latent factors. To estimate the states, we adopt Hidden Markov Model (HMM). Moreover, we use Hotelling's theory to detect abnormal data statistically. By using our approach, even if states can not be explicitly obtained from a device, hidden states can be estimated to perform anomaly detection in more details.

With the rapid development of information and communication technology, a vast amount of personal health data is generated, stored, and utilized in healthcare related services. However, there are many issues remained to be solved, such as data interoperability, security, and privacy concerns. On the other hand, blockchain, a decentralized peer-to-peer digital ledger, has attracted a lot of attention in recent years as a promising technology to protect privacy of personal data. By adopting blockchain, individuals can take advantage of personal health data for better healthcare. In this paper, after briefly summarizing the major features of blockchain, we describe blockchain-empowered solutions for utilization of personal health data in healthcare and discuss issues and challenges. We further propose i-Blockchain, an individual-centric framework for use of personal health data based on an extension of permissioned blockchain, present the basic architecture and protocols, and show several application scenarios.

It has become possible to record the data of a visited place automatically, using a smartphone's GPS. The smartphone can visualize where a user visits and present the traveling route. In this study, we propose a personal whereabouts model for living pattern analysis, utilizing the smartphone's GPS. The proposed model aims to extract personal living patterns effectively. To this propose, it is important to analyze those factors related to location and environment. A decision tree based method is used to extract these factors for the location-based personal data analysis. We finally discuss the experiment design and our observations based on personal data collected from mobile environments.

In recent years, it is very popular to use Twitter as an interaction tool and for information transmission and sharing as well. It is more difficult for us to understand the interaction flow on Twitter than the usual dialogue and conversation. But it will become easier to catch the interaction flow if we can focus on the priority words of Tweets. As we know, the word priority of Tweets can be measured by the betweenness centrality based on the word co-occurrence network. However, it is unknown whether the word priority measured by the betweenness centrality is the same to what the Twitter users think or not, which has not yet been well studied. In this study, we design an experiment to compare the measurement calculated by the betweenness centrality and the evaluation given by subjects. We further analyze their difference by statistical methods. Our result shows that the measurement determined by the betweenness centrality does not always match the subjective evaluation, but they tend to have a positive correlation. Particularly, proper nouns and words with long characters are more likely to be regarded higher priority than the betweenness centrality.

Personal data is data related to an individual, generated by an individual, or metadata about an individual. To analyze personal data comprehensively, it is needed to consider different types and sources of data. Moreover, it should be considered not only explicit attributes but also latent factors. In this study, to specify latent factors, we use Structural Equation Modeling (SEM) with a domain model for personal data analysis. The domain model represents the relationship between the latent factors and measures that are possible to be obtained by a wearable device. We construct an activeness model as the domain model and apply it for personal data analysis. The activeness level which is assumed as the latent factor is quantified by SEM. We verify the adaptability of the activeness model by comparing the case of classifying by the activeness factor with the case of not using latent factors. The result shows that the model has higher adaptability when personal data is classified by latent factors than only by labels.

In this study, we use wearable devices to monitor the health and physiological indicators of the elderly, and analyze the effects of the elderly's walking steps and sleep as the measurable health and physiological indexes. Eighteen cases of the elderly who received health management services are selected. We establish the generalized linear mixed model, and Pearson correlation analysis result shows statistically significant for the number of walking steps and systolic blood pressure/diastolic blood pressure. The number of walking steps is positively correlated with the systolic/diastolic pressure. The more walking steps and the better the sleep quality, the more stable the daily blood pressure is. Therefore, it is possible to effectively control the elderly's blood pressure by exercising and managing sleeping. It can be better for the elderly to have self-esteem health managed by using wearable devices, and encourage the elderly to exercise at a certain extent.

In recent years, recommender systems are widely applied in e-commerce system to help users locating their interested information. However, the 'all good reputation' problem brings down the accuracy of recommender systems. In addition, users' social network can benefit the recommendation especially when dealing with cold-start scenarios. In this paper, a novel trust-aware recommendation approach for e-commerce is proposed, which unearths the hint from ordinary rating and trust network by users' instant interactions in e-commerce system. More precisely, a rating revamping algorithm is designed to extract semantic ratings from feedback comments, and further construct fine grained rating score for the next process. Then, the recommendation scheme is studied through analyzing the users' trust network and their own behavior in e-commerce system. Finally, evaluations conducted based on a real dataset 'Douban' to demonstrate the effectiveness of the proposed method.

As an SNS, Twitter is popular because users can post their emotions as a short message easily. Emotional tweets may influence user relationships. In our previous study, we found that positive users construct mutual relationships in Twitter. Keyword matching with emotional word dictionaries was used to detect positive users. The problem of keyword matching is the limitation of word number. To solve this problem, we use machine learning, specifically Naive Bayes Classification, to classify emotions of tweets. We analyze whether there is a difference in user relationships between the positive and negative groups by the Brunner-Munzel test. The result shows that the relationships of positive users increase more than that of negative users in the followee fluctuation, follower fluctuation and mutual follow fluctuation, which means that a positive user is more active to construct user relationships than a negative user.

Data intensive computing (DIC) provides a high performance computing approach to process large volume of data. In this study, a new formalization is introduced to present the two-stage DIC task execution in a stream manner. A novel heuristic algorithm is proposed for the scheduling problem due to the NP complexity. The theoretical approximation ratio bounds for the heuristic are analyzed and confirmed by the experimental evaluation. Overall, we observe that the proposed method conducts average 1.2 times makespan than the theoretic bound of the optimal solution. Besides, the proposed method outperforms the GA and FIFO scheduling schemes with overall improvements. (C) 2017 Elsevier Inc. All rights reserved.

In our society, inadequate physical activity is one of severe problems issues for human health, which may increase the health risks of many diseases. Nowadays, smartphones are ubiquitous and widely used around the world, in which multi-functional sensors and wireless interfaces are embedded. Therefore, smartphone is viewed as an appropriate platform for real-time activity recognition to address these healthy problems by monitoring and detecting user's everyday activities. In this paper, unlike other wearable devices based applications (e.g., watches, bands, or clip-on devices), ActiRecognizer, a smartphone-based prototype of a real-time human activity recognition (HAR) is designed and implemented, in which a detailed activity report of individuals (i.e. a pie chart containing the proportion and duration of each activity) can be correspondingly generated based on the detected real-time activities. Specifically, ActiRecognizer adopts client/server (C/S) architecture. At client side, smartphone associated with each individual periodically uploads the accelerometer and gyroscope sensing data to server for activity recognition and monitoring. At serve side, HAR is composed of offline training phase and online activity recognition phase: In training phase, sensing data are collected to extract the desired features that can appropriately characterize behaviors, classification model is generated utilizing these features, and then the trained classification model is used to classify user activity in real time. Finally, detailed activity reports and statistics are available to the user via a secure web interface.

In recent years, cross-disciplinary scientific collaboration has been proved to be promising for both research practice and innovation. Lots of efforts have been spent in collaboration recommendation. However, the cross-disciplinary information is hidden in tons of publications, and the relationships between different fields are complicated, which make it challengeable recommending cross-disciplinary collaboration for a specific researcher. In this paper, a novel cross-disciplinary collaboration recommendation method (CDCR) that unearths the common cross-disciplinary collaboration patterns and historical scientific field preferences of authors is proposed to recommend potential cross-disciplinary research collaboration. In CDCR, a research field discovery algorithm is designed to classify scientific topics obtained from the publications into the correct field automatically. Then, the collaborative patterns are studied through analyzing the composition fields and the corresponding percentage of all publications. Furthermore, we investigate the common correlation of different research fields. Based on the common correlation and the researcher's specific pattern, the most valuable fields will be listed by CDCR. The effectiveness of our approach is evaluated based on a real academic dataset.

Crowdsourcing systems which rely on a great deal of crowdworkers to perform large quantities of microtasks, have been leveraged in a variety of applications. There are two factors affecting the productive output of each crowdworker. One is skill level, which is private information to each crowdworker, and another is her variable expended effort. In this paper, we construct and analyze a total-Ability-balanced team based incentive mechanism ABT, which can stimulate the strategic crowdworkers to truthfully report their ability levels, and according to crowdworkers' ability levels, form the competing teams. Specifically, a crowdworker with a certain skill level, is askedto choose a specificskill level (i.e., denoted as an ability threshold), and a basic payment scheme is designed to incentivize the crowdworker to truthfully report her ability level. Then, according to the chosen ability thresholds, crowdworkers are organized into total ability balanced teams to earn extra team bonus, which can further motivate crowdworkers to exert more efforts. Compared to team formation process where workers are randomly assigned to the same-scale teams and the pay per task model, our scheme ABT can improve the work efficiency.

Recently, Device to Device (D2D) based mobile social networking in proximity (MSNP) has witnessed great development on smartphones. Unlike traditional OTT (Over-The-Top) based proximal social services (like Foursquare, etc.), in which the centralized server (usually located in the cloud) receives periodic location updates from users' mobile devices (e.g., using GPS, etc.), and then determines proximity based on gathered locations and interests, D2D based proximity services (MSNPs) support infrastructure-free and self-organized social networking, allowing users to share experiences in real time and in a specific place, which are more fun, and can significantly increase users' engagement. However, the glaring absence of the practical MSNP solutions on the market is alarming, due to the lack of comprehensive understanding of architecture, technologies and development frameworks available (advantages/disadvantage, applicability, etc.), as well as special requirements of MSNPs. This paper thoroughly investigates D2D based MSNP architecture, key challenges and their potential solutions on smartphones, organized as layers from bottom to up, presents how to utilize commercial off-the-shelf existing technologies and development frameworks to quickly and easily build MSNP applications, and summarizes the unsolved issues in popular development frameworks (e.g., AllJoyn, etc.). Our primary goal is, for commercially available smartphones, to identify key challenges in various layers of MSNP architecture, potential development frameworks and solutions, provide quick how-to-do for developers of MSNP applications, and greatly facilitate to develop and deploy interesting and commercially potential MSNP applications. (C) 2015 Elsevier B.V. All rights reserved.

In the DNA microarray research field, the increasing sample size and feature dimension of the gene expression data prompt the development of an efficient and robust feature selection algorithm for gene expression data classification. In this study, we propose a hybrid feature selection algorithm that combines the mutual information maximization (MIM) and the adaptive genetic algorithm (AGA). Experimental results show that the proposing MIMAGA-Selection method significantly reduces the dimension of gene expression data and removes the redundancies for classification. The reduced gene expression dataset provides highest classification accuracy compared to conventional feature selection algorithms. We also apply four different classifiers to the reduced dataset to demonstrate the robustness of the proposed MIMAGA-Selection algorithm. (C) 2017 Elsevier B.V. All rights reserved.

With the rapid development of mobile networks, people are no longer limited to seeking information offline. The Internet provides a good choice for them. As the previous research work shows, more and more people get used to using the Internet to search for health information. Online health information has attracted much attention ever before. It is well known that health related information generated on the Internet contains a large amount of practical data, which can be utilized for health related study. In this paper, firstly we review related research work on self-disclosure and online behavior related to health information. We analyze and summarize factors relevant to health concerns, and discuss measuring dimensions for health concerns. Finally, we present a basic health concern model and future direction for health concern analysis based on online health information behaviors and social data.

The development of social media provides convenience to people's lives. People's social relationship and influence on each other is an important factor in a variety of social activities. It is obviously important for the recommendation, while social relationship and user influence are rarely taken into account in traditional recommendation algorithms. In this paper, we propose a new approach to personalized recommendation on social media in order to make use of such a kind of information, and introduce and define a set of new measures to evaluate trust and influence based on users' social relationship and rating information. We develop a social recommendation algorithm based on modeling of users' social trust and influence combined with collaborative filtering. The optimal linear relation between them will be reached by the proposed method, because the importance of users' social trust and influence varies with the data. Our experimental results show that the proposed algorithm outperforms traditional recommendation in terms of recommendation accuracy and stability.

Recently, with the widespread popularity of SNS (Social Network Service), such as Twitter, Facebook, people are increasingly accustomed to sharing feeling, experience and knowledge with each other on Internet. The high accessibility of these web sites has allowed the information to be spread across the social media more quickly and widely, which leads to more and more populations being engaged into this so-called social stream environment. All these make the organization of user relationships become increasingly important and necessary. In this study, we try to discover the potential and dynamical user correlations using those organized social streams in accordance with users' current interests and needs, in order to assist the collaborative information seeking process. We develop a heuristic approach to build a Dynamically Socialized User Networking (DSUN) model, and define a set of measures (such as interest degree, and popularity degree) and concepts (such as complementary tie, weak tie, and strong tie), to discover and represent users' current profiling and dynamical correlations. The corresponding algorithms are developed respectively. Finally, the architecture of the functional modules is presented, and the experiment results are demonstrated and discussed based on an application of the proposed model.

Location-based social networks (LBSNs) have recently attracted millions of mobile users to share their locations and location-related contents. With the increasing use of LBSNs, an efficient personalized recommendation service is required to recommend appropriate point of interests (POIs) to users. Traditional collaborative filtering (CF) based recommendation algorithms need go through all users in LBSN to recommend locations to the target user. Due to the fact that many users are irrelevant to the target user, these approaches perform poorly in accuracy and scalability. In this paper, we propose an Efficient Location Recommendation scheme based on Discrete particle swarm optimization (DPSO) and Collaborative filtering, called ELR-DC. This scheme efficiently detects communities with close internal ties and then conducts location recommendation in each community. Specifically, a similarity network among users is firstly constructed based on their check-in activities, which explicitly takes into account users' similarities of interest and active regions. Then, an improved merging DPSO algorithm (IMDPSO) is proposed to detect communities through utilizing the formed similarity network. Then, in each community, CF algorithm is applied to recommend Top-N locations to each user. Finally, we conduct a comprehensive performance evaluation on a large-scale datasets collected from Gowalla. Experimental results show that the proposed scheme have the superiority of the precision and efficiency over the existed CF algorithms.

Falling accidents cause severe damage to human health, especially to the elderly. Nowadays, smartphones are ubiquitous and widely used around the world, in which rich sensors are embedded. Thus, a smartphone based falling detection system, SmartFDS is proposed in this paper, which can recognize the associated individual's falling behavior and send out an emergent message containing the location for immediate help. Generally, SmartFDS includes offline training phase and online activity recognition phase. Specifically, in training phase, utilizing the embedded accelerometer sensor, falling data for various situations are collected to extract the desired features that can appropriately characterize the falling behavior. Considering that the raw data are intrinsically prone to noise and error, those data are preprocessed by weighted smoothing. Then, 6 time-domain features are elaborated to determine the discriminative features for characterizing falling, and two features, maximum and vertical velocity are selected. Then, 4 different classification algorithms are investigated, and SVM is selected. The prototype of SmartFDS is implemented on Android phones, and can detect falling in real time accurately.

Crowdsourcing is the generalized act of outsourcing tasks, traditionally performed by an employee or contractor, to a large group of Internet population through an open call. With the great development of smartphones with rich built-in sensors and multiple ratio interfaces, mixing smartphone-based mobile technologies and crowdsourcing offers significant flexibilities and leads to a new paradigm called mobile crowdsourcing (MCS), which can be fully explored for real-time and location-sensitive crowdsourced tasks. In this paper, we present a taxonomy for the MCS applications, which are explicitly divided as using human as sensors, and exploiting the wisdom of crowd (i.e., human intelligence). Moreover, two paradigms for mobilizing users in MCS are outlined: direct mode and word of mouth mode. A comprehensive MCS framework and typical workflow of MCS applications are proposed, which consist of nine functional modules, pertaining to three stakeholders in MCS: crowdsourcer, crowdworkers, and crowdsourcing platform. Then, we elaborate the MCS challenges including task management, incentives, security and privacy, and quality control, and summarize the corresponding solutions. Especially, from the viewpoints of various stakeholders, we propose the desired properties that an ideal MCS system should satisfy. The primary goal of this paper is to comprehensively classify and provide a summary on MCS framework, challenges, and possible solutions to highlight the MCS related research topics and facilitate to develop and deploy interesting MCS applications. Copyright (C) 2016 John Wiley & Sons, Ltd.

Scientific research works conducted by researchers spread all over the world in every research field, which are hard to be tracked and quantified. Although there are many research works focused on scientific community discovery and researcher profiling, it is still a big challenge to track the research patterns and assess the research development for an individual researcher or a research group over time. In this study, we seek to model researchers' scientific activities and quantify their outcome during their research career. A temporal tracking and assessing model is introduced to represent the research development and quantify the scientific outcome for both an individual and a group along the time. Based on our model, a research topic analyzing approach is developed to extract the topics covered by a research group for the research pattern analysis. Furthermore, a latent research pattern discovering approach is proposed to depict how a research group's research works contributed by its members are discovered and visualized. The effectiveness of our approach is evaluated based on a real academic dataset.

Today's smartphones with a rich set of cheap powerful embedded sensors can offer a variety of novel and efficient ways to opportunistically collect data, and enable numerous mobile crowdsourced sensing (MCS) applications. Basically, incentive is one of fundamental issues in MCS. Through appropriately integrating three popular incentive methods: reverse auction, reputation and gamification, this paper proposes a quality-aware incentive framework for MCS, QuaCentive, which, pertaining to all components in MCS, can motivate crowd to provide high-quality sensed contents, stimulate crowdsourcers to give truthful feedback about quality of sensed contents, and make platform profitable. Specifically, first, we utilize the reverse auction and reputation mechanisms to incentivize crowd to truthfully bid for sensing tasks, and then provide high-quality sensed contents. Second, in to encourage crowdsourcers to provide truthful feedbacks about quality of sensed data, in QuaCentive, the verification of those feedbacks are crowdsourced in gamification way. Finally, we theoretically illustrate that QuaCentive satisfies the following properties: individual rationality, cost-truthfulness for crowd, feedback-truthfulness for crowdsourcers, platform profitability.

Research trends keep evolving along the time with certain trackable patterns. Mining academic literature and discovering the latent research trends evolution is an interesting and important problem. Few of previous studies focusing on academic topic evolution modeling have addressed the temporal topic evolution patterns. In addition, researchers' profile and their social networks are valuable complementary to the research trends tracking. In this study, to analyze the underlying research trends evolution along with the scientific collaborations of researchers, a novel temporal research trends evolution model associated with researchers' social networks is proposed and built. Specifically, the detected research topics are classified into different clusters in each timeslot, and the evolution patterns are deduced among these topic clusters. The effectiveness of our approach is evaluated based on a real academic dataset. The experimental results can help users to discover the major research trends for specific fields. Besides, the tracked statuses of the corresponding scientific groups are helpful for searching research trends or finding collaboration opportunities according to researchers' different requirements.

In recent years, various location based services (LBS) have witnessed great development and are being prevalently used in our life. However, as the foundation of various LBS applications, localization consumes large energy of resource-constraint mobile terminals, especially on smartphones. This paper explicitly proposes three technical principles, substitution, adaption and collaboration to guide energy-efficient localization and tracking schemes on smartphones. Then several typical schemes in indoor or outdoor environments are respectively summarized and compared under the umbrella of those three principles. Moreover, the context-Assisting techniques are also discussed to design energy-efficient LBS applications. Finally, the quantitative metrics to measure the tradeoff between energy and localization performance are summarized. The primary goal of this paper is to comprehensively classify and provide a summary on the sporadic localization schemes (with energy-efficiency as main concern), possible solutions and tradeoffs, and facilitate to develop and deploy the energy-efficient LBS applications.

Currently, mobile devices (e.g., smartphones) are equipped with multiple wireless interfaces and rich built-in functional sensors that possess powerful computation and communication capabilities, and enable numerous Mobile Crowdsourced Sensing (MCS) applications. Generally, an MCS system is composed of three components: a publisher of sensing tasks, crowd participants who complete the crowdsourced tasks for some kinds of rewards, and the crowdsourcing platform that facilitates the interaction between publishers and crowd participants. Incentives are a fundamental issue in MCS. This paper proposes an integrated incentive framework for MCS, which appropriately utilizes three widely used incentive methods: reverse auction, gamification, and reputation updating. Firstly, a reverse-auction-based two-round participant selection mechanism is proposed to incentivize crowds to actively participate and provide high-quality sensing data. Secondly, in order to avoid untruthful publisher feedback about sensing-data quality, a gamification-based verification mechanism is designed to evaluate the truthfulness of the publisher's feedback. Finally, the platform updates the reputation of both participants and publishers based on their corresponding behaviors. This integrated incentive mechanism can motivate participants to provide high-quality sensed contents, stimulate publishers to give truthful feedback, and make the platform profitable.

SUMMARY In this paper, we propose a gradually adaptive recommendation model based on the combination of both users' commonalities and individualities that depend on the semantic mapping of users' interest correlations. We analyze users' information access behaviors and histories to extract users' interests and trace their transitions. In details, according to a set of bookmark tags classified by a semantic means, the pages accessed by users are assigned into several tag classes, which will finally be clustered into different groups in accordance with the types of interests that belong to two categories: personal and common interests, respectively. Based on the detection of users' interest focus transitions through interactions between users, we provide a series of information seeking actions in sequence to the target users. Besides, according to the reference groups which are defined to describe different relations with the target users, the successful experience is extracted and recommended. After the description of the definitions and measures, the mechanism to infer the interest focus, the system architecture and experimental evaluation results are described and demonstrated.

In this paper, we propose a gradually adaptive recommendation model based on the combination of both users' commonalities and individualities that depend on the semantic mapping of users' interest correlations. We analyze users' information access behaviors and histories to extract users' interests and trace their transitions. In details, according to a set of bookmark tags classified by a semantic means, the pages accessed by users are assigned into several tag classes, which will finally be clustered into different groups in accordance with the types of interests that belong to two categories: personal and common interests, respectively. Based on the detection of users' interest focus transitions through interactions between users, we provide a series of information seeking actions in sequence to the target users. Besides, according to the reference groups which are defined to describe different relations with the target users, the successful experience is extracted and recommended. After the description of the definitions and measures, the mechanism to infer the interest focus, the system architecture and experimental evaluation results are described and demonstrated. Copyright (c) 2014 John Wiley & Sons, Ltd.

Nowadays, we are living in a well-suited social environment with a variety of lifestyles and values. Life support has become important in such a diversified society. Along with continuously collecting the tremendous amount of personal big data generated in the social environment, it is possible for us to provide the life support based on personal data analytics. Moreover, analyzing such a kind of data can facilitate deep understanding of individual life. In this study, we focus on personal data analytics to support well-being oriented life. Three categories of personal data are classified from the collection of individuals' daily life data, and a framework of well-being oriented personal data analysis is proposed, which can provide people with suggestions and advices to improve their living life.

Cyberization, as a new big trend following computerization and informatization, is the process of forming a new cyberworld and transforming our current physical, social and mental worlds into novel cyber-combined worlds. Cyber science, responding to the cyberization trend, aims to create a new collection of knowledge about these cyber-enabled worlds, and provide a way of discovering what is in the cyber-enabled worlds and how they work. Cyber science is concerned with the study of phenomena caused or generated by the cyberworld and cyberphysical, cyber-social and cyber-mental worlds, as well as the complex intertwined integration of cyber physical, social and mental worlds. It fuels advances in cyber technology, beyond the existing cyber related technologies. In this paper, after discussing the cyberization background and process, and explaining cyber science and technology, we present our visions and perspectives on new opportunities, essential issues and major challenges for cyber science and technology. We further describe cyber related technologies and closely related existing research areas, and envision future research directions, in terms of cyber physical, cyber social, cyber life, cyber intelligence and cyber security, which are five basic dimensions of cyber science and technology.

The high development of emerging computing paradigms, such as Ubiquitous Computing, Mobile Computing, and Social Computing, has brought us a big change from all walks of our work, life, learning and entertainment. Especially, with the high accessibility of social networking services along with the increasingly pervasive use of portable wireless mobile computing devices, more and more populations have been engaged into this kind of integration of real physical world and cyber digital space, which can be called the hyper world. To help people live better in the highly developed information society, the so-called cyber-individual (Cyber-I), which is far beyond a user model or a software agent to assist a user, has been proposed to provide the most comprehensive digital entities for its corresponding Real-I in terms of the individual's experience, behavior, and thinking as well as his or her birth, growth, and death. In this study, we concentrate on the personal data analytics to facilitate the cyber individual modeling. Organic Stream is introduced to systematically organize and refine the personal stream data, which can help improve the data processing and management in the CI-Spine tier and CI-Pivot tier of Cyber-I. The DSUN (Dynamically Socialized User Networking) model is employed to better utilize the collective intelligence from a group of users, which can help improve the CI-Mind tier to make Cyber-I to become more robust. Based on these, we discuss the functional modules for the facilitation of cyber individual modeling. Finally, a scenario is given, and the experimental results are presented to demonstrate that the valuable outcomes from the personal analysis can be utilized to enrich the Cyber-I, and provide users with more suitable services.

Device to Device (D2D) based mobile social networking in proximity (MSNP) has witnessed great development on smartphones. Message forwarding/data routing in MSNP relies on the movement of individuals and their encounter opportunities to relay data to the destination, so-called store-carry-forward (SCF) mode. Due to lacking continuous network connectivity and must handle dynamic topology, message forwarding mechanism in MSNP is challenging. In recent years, exploiting MSNP users' social behaviours to forward message, have drawn tremendous interests. In this paper, we provide the taxonomy of social behaviour-based forwarding strategies, comprehensively summarize and compare various schemes, including location-based and encounter-based (further divided into social property based and community-based) strategies. Location-based strategies forward data to the individuals geographically closer to the destination, while encounter-based strategies forward data to the users more socially active through considering human interactions. Then, we present open issues in MSNP forwarding strategies and their potential solutions. The goal of this work is to provide deep understanding on SCF based message forwarding in MSNP which properly incorporate the individual's social characteristics, and bring new visions to MSNP research and applications.

Community structure is a common feature in real-world network. Overlap community detection is an important method to analyze topology structure and function of the network. Most algorithms are based on the network structure, without considering the node attributes. In this paper, we propose an overlapping community detection algorithm based on node convergence degree which combines the network topology with the node attributes. In our method, PageRank algorithm is used to get the importance of each node in the global network and utilize the local network (local neighbors) to measure the structure convergence degree. Then, node convergence degree combining node attributes and structure convergence degree is designed. Finally, the overlap communities can be identified by the Spectral Cluster based on node convergence degree. Experiments results demonstrate effectiveness and better performance of our method.

In this study, we concentrate on a feasibility study on personal data analysis for well-being oriented life support. We introduce our basic concept and model to describe the personal data collected from people's daily life, and discuss how to utilize the personal analysis to provide users with the individualized services in their daily lives. Finally, we present the experimental analysis based on people's daily activity data, to demonstrate the feasibility of our proposed approach.

The world has become an aging society with exponentially increasing social security benefit expenditure and nursing care costs. To help solve these issues and challenges, utilization of advanced ICT is highly expected to allow more elderly people to remain independent for their proactive social participation irrespective of age. In this study, we propose a human-centric safe and secure framework of ubiquitous living environments (Ubi-Liven) for the elderly people towards seamless integration of the cyber-enabled ubiquitous holistic living support system with a physical living environment. We further address and discuss the design and technical issues for the implementation of a smart living environment on the fly under the proposed framework, empowered by ubiquitous assistive technologies such as cloud, IoT and big data analytics based on life logs to provide holistic support for the elderly's activities of daily living and healthcare.

Recent years has witnessed a great progress in Vehicular Social Networks (VSNs), which is a particular class of vehicular ad hoc networks, characterized by social aspects and features, including human mobility and rational behaviors. Among various VSN applications, one of the most promising applications is the dissemination of commercial advertisement via car-to-RSU (rode side unit) and car-to-car communication. However, due to non-cooperative behavior of selfish nodes or malicious ones in the real-world scenario, such vehicular advertisement system can't be realized unless proper security and incentive mechanisms are taken into account. In this paper, the privacy concern is addressed by leveraging a PKI (Public Key Infrastructure) to provide secure incentives for cooperative nodes. The selfishness issue is solved through adopting Topology dependent Reward Mechanisms (TDRM). Preliminary theoretical and simulation results show that, besides incentivizing individuals' participation, the proposed scheme can increase participants' reward through their active contribution and solicitation, and moreover, is robust against users' misbehaviors, especially Sybil-attack.

Nowadays, the analysis of social networks, as well as the community evolution has become a hotly discussed topic in social computing field. In this paper, we focus on mining and tracking the dynamic communities based on social networking analysis. Based on a generic framework for the dynamic community discovery, a computational approach is developed to extract users' static and dynamic features for the temporal trend detection. A dynamically socialized user networking model is then presented to describe users' various social relationships. A mechanism is proposed and developed to detect the dynamic user communities, and track their evolving changes. Experiments using Twitter data demonstrate the effectiveness of our method in tracking how communities dynamically create, split, and merge from a group of connected people in social media environments.

Considering that modern mobile terminals possess the capability to detect users' proximity, and offer means to directly communicate and share content with the people in close area, Device-to-Device (D2D) based Proximity Services (Pro Se) have recently witnessed great development, which enable users to seek for and utilize relevant value in their physical proximity, and are capable to create numerous new mobile service opportunities. However, without a breakthrough in battery technology, the energy will be the biggest limitation for ProSe. Through incorporating the features of ProSe (D2D communication technologies, abundant built-in sensors, localization-dependent, and context-aware, etc.), this paper thoroughly investigates the energy-efficient architecture and technologies for ProSe from the following four aspects: underlying networking technology, localization, application and architecture features, context-aware and user interactions. Besides exploring specific energy-efficient schemes pertaining to each aspect, this paper offers a perspective for research and applications. In brief, through classifying, summarizing and optimizing the multiple efforts on studying, modeling and reducing energy consumption for ProSe on mobile devices, the paper would provide guide for developers to build energy-efficient ProSe.

This article focused on seeking a new heuristic algorithm for the influence maximization problem in complex social networks, in which a small subset of individuals are intentionally selected as seeds to trigger a large cascade of further adoptions of a new behavior under certain influence cascade models. In literature, degree and other centrality-based heuristics are commonly used to estimate the influential power of individuals in social networks. The major issues with degree-based heuristics are twofold. First, those results are only derived for the uniform IC model, in which propagation probabilities on all social links are set as same, which is rarely the case in reality; Second, intuitively, an individual's influence power depends not only on the number of direct friends, but also relates to kinds of those friends, that is, the neighbors' influence should also be taken into account when measuring one's influential power. Based on the general weighted cascade model (WC), this article proposes Pagerank-inspired heuristic scheme, PRDiscount, which explicitly discounts the influence power of those individuals who have social relationships with chosen seeds, to alleviate the "overlapping effect" occurred in behavior diffusion. Then, the authors use both the artificially constructed social network graphs (with the features of power-law degree distribution and small-world characteristics) and the real-data traces of social networks to verify the performance of their proposal. Simulations illustrate that PRDiscount can advantage over the existing degree-based discount algorithm, Degree Discount, and achieve the comparable performance as greedy algorithm.

With the development of social network services, the user relation spectrum of the social network has exceeded our imagination. Hence, personalized recommendation algorithms are adopted in many social networking sites to help users find their potential friends and related information more quickly and conveniently. In this paper, we discuss the weaknesses of current algorithms, and propose a user profile integrated dynamic social recommendation algorithm in order to overcome those limitations. Finally, through the experiment on Weibo dataset, it can conclude that the proposed algorithm outperforms traditional approaches in terms of accuracy and stability.

In this study, we concentrate on analyzing and building the dynamical user profiling to describe users' multi-dimensional features and properties, in order to assist the individualized information seeking and recommendation process in social networking environments. A set of user attributes are introduced and defined to describe the basic user profiling in accordance with the analysis of information behaviors, and several centrality based measures are proposed and developed to describe the users' importance and contributions with regards to a group of users based on their social connections in the DSUN (Dynamically Socialized User Networking) model. The experimental results are discussed to demonstrate the feasibility and effectiveness of our proposed methods.

With the increasing popularity of social network services, a tremendous amount of information has been produced. And with more and more users involving in the SNS environment, their network relationships have become as complex as in the real world. In order to find out the useful information more efficiently, in this study, we propose a participatory search and recommendation system based on users' social roles and their relationships in the SNS environment. The proposed system is used to filter users and their messages in a participatory way by analyzing their social roles and connection networks between each user, which can further contribute to personalized information search and recommendation. We describe the design and implementation issues of a prototype system, and discuss how to use the social roles and connection networks to support and empower information search and recommendation.

Collaborative filtering (CF) is a popular method for the personalized recommendation. Almost all of the existing CF methods rely only on the rating data while ignoring some important implicit information in non-rating properties for users and items, which has a significant impact on the preference. In this study, considering that the average rating of users and items has a certain stability, we firstly propose a personalized fitting pattern to predict missing ratings based on the similarity score set, which combines both the user-based and item-based CF. In order to further reduce the prediction error, we use the non-rating attributes, such as a user's age, gender and occupation, and an item's release date and price. Moreover, we present the deviation adjustment method based on the support vector regression. Experimental results on MovieLens dataset show that our proposed algorithms can increase the accuracy of recommendation versus the traditional CF.

Recent advances in information technology have turned out World Wide Web to be the main platform for interactions where participants-users and corresponding events-are triggered. Although the participants vary in accordance with scenarios, a considerable size of data will be generated. This phenomenon indeed causes the complexity in information retrieval, management, and resuse, and meanwhile, turns down the value of this data. In this research, we attempt to achieve efficient management of user-generated data and its derivative contexts (i.e., social ad hoc data) for human supports. The correlations among data, contexts, and their hybridization are specifically concentrated. An intelligent state machine is proposed to outline the relations of data and contexts, and applied to further identify their usage scenarios. The performance and feasibility can be revealed by the experiments that were conducted on the data collected from open social networks (e.g., Facebook, Twitter, etc.) in the past few years with size around 500 users and 8,000,000 shared contents from them.

In recent years, mainly driven by the availability of modern mobile devices with cheap integrated position sensors, location based services (LBS) have become more and more popular. Prominent examples are nearest friends finding in mobile social networks and the points of interest finders such as the nearest gas stations, hospitals, or places of interests etc. However, the users' privacy information such as location is threatened, when users enjoy the convenience and effectiveness provided by such location services. Therefore, how to secure users' privacy information must be taken into consideration. Many different approaches have been proposed to protect users' identity, location and so on. This paper reviews and analyzes existing privacy protection research works from an integrated perspective, gives the LBS category from two views, and discusses the challenges of securing privacy information. Lastly, our suggestions for future research works are presented.

Recently, Device to Device (D2D) based mobile social networking in proximity (MSNP) has witnessed great development on smartphones, which enable actively/passively and continuously seek for relevant value in one's physical proximity, through direct communicating with other individuals within the communication range, without the support of centralized networking infrastructure. Specially, a user would like to find out and interact with some strangers with similar interest in vicinity through profile matching. However, in matching process, individuals always have to reveal their personal and private profiles to strangers, which conflicts with users' growing privacy concerns. To achieve privacy preserving profile matching (i.e., friend discovery), many schemes are proposed based on homomorphic and commutative encryption, which bring tremendous computation and communication overheads, and are not practical for the resource limited mobile devices in MSNP. In this paper we adapt Confusion Matrix Transformation (CMT) method to design a Lightweighted fIne-grained Privacy-Preserving Profile matching mechanism, LIP3, which can not only efficiently realize privacy-preserving profile matching, but obtain the strict measurement of cosine similarity between individuals, while other existing CMT-based schemes can only roughly estimate the matching value.

Micro-Learning is a new leaning paradigm based on microblogging, e-mail or SMS, in which an integral learning resource consists of a series of micro learning units that are dispersed via the services of Internet, and can be used to help users learn at anywhere with a short-term. But the problem is that most of the micro learning units are disorganized. Therefore, it is a critical issue how to organize these micro learning units, in order to make them easier to be used and learned. In this study, we propose an approach based on process mining to organize the learning units according to the situation of users. Firstly, the successful micro-learning processes are extracted from the access logs. And then, the learning process map named navigation map is created based on the domain knowledge and the access sequence of users. Secondly, according to the similarity of access behavior, a reference user group is extracted dynamically for a target user. Based on the dynamic Bayesian network, the navigation map is then used to calculate the posterior probabilities of learning units. Finally, on the basis of posterior probabilities, a learning unit can be recommended for the target user to learn as the next step. The results are provided to the target user gradually, until the target user finish the whole course by the guide of learning process at his/her fragmented time.

With the popularity of mobile Internet, smart phones and location-based services, Mobile Social Network in Proximity (MSNP) has attracted more and more attentions, for which trustworthy service discovery and its associated latency are two important issues. In this paper, after overviewing MSNP and trustworthiness for service discovery in MSNP, we discuss the major issues and potential solutions of trustworthy service discovery. We present trustworthiness determination strategies based on service provider candidates' past experience and current profile analysis to reduce the latency caused by trustworthiness computation for trustworthy service discovery in MSNP.

Recently, mobile social networks in proximity (MSNP) has gained tremendous attentions, which explicitly explores the physical proximity of individuals, and greatly facilitates their face-to-face interactions through direct communications. Generally, Proximity awareness can be enabled with two methodologies: Over-The-Top (OTT), in which the centralized server determines the proximity based on users' location updates and interests, and intermediate the interactions between proximal users; Device-to-Device (D2D), which allows two devices in radio range to discovery and communicate directly. This paper designs and implements a novel hybrid interactive chatting system, HYChat, which integrates the merits of both methodologies: simultaneously support direct interaction through WiFi Direct when individuals are within each other's radio range; when detecting two devices are out of range of WiFi Direct (e.g., due to individual's movement), it will switch to online OTT mode automatically to continue the interactive chat. Specifically, two key challenges in HYChat are preliminarily solved in this paper: profile matching prior to establishing connection among devices (i.e., layer 2 profile matching), and the smooth switching between D2D and OTT modes. The prototype of HYChat on Android phones illustrates its effectiveness. In brief, the hybrid system HYChat could greatly facilitate individuals' social interaction in physical proximity.

Although there are many research works focused on scientific community discovery and researcher profiling, it is still a challenge to track and assess the research development for an individual researcher or a research group over time. In this study, we seek to model and quantify an individual's research work outcome during his/her research career. A temporal tracking model is introduced to represent research development for both an individual and a group along the time. Furthermore, assessing measures are proposed to depict how to quantify a research group's outcomes by its members' contribution. Based on our model, a research topic analyzing approach is proposed to extract the topics covered by a group and discover the group's research pattern. The analyzed output is helpful for a researcher to pursue his/her scientific life or find collaboration opportunities.

With the help of several emerging computing paradigms, such as Ubiquitous Computing, Social Computing, and Mobile Computing, OLP (Open Learning Process) has become an important research issue in the online learning environment. It becomes possible to share the learning process related information in and cross systems to meet the needs of all stakeholders, such as teachers, learners and managers. In this study, we focus on the OLP in the social learning environment based on the personal and social analytics, which can provide users with the process-oriented learning support. We present a data processing framework to collect, analyze, and organize the learning-related data in a user-centric way. Three important issues regarding to the unified data management, socialized learning analytics, and sharable open learning process are discussed towards the process-oriented learning innovation. Based on these, an open learning platform is proposed and designed to provide users with the individualized support in the collaborative learning process.

Understanding research topic evolution in a specific field is important to learn the temporal structure and latent research trends. Although research works in terms of topic analysis have been developed for many years, it is still difficult to well understand the topic evolution due to the limitation of the bibliographical data which does not consider the researchers' profiles and their social networks. In this study, to analyze the underlying research topic evolution and facilitate the scientific collaborations for researchers, a novel topic evolution model based on the analysis of researchers' social networks is built. Specifically, the researchers' co-authorships associated with their latent social networks are analyzed. A research topic tracking method is then developed to track the topic evolutions and analyze the research lifecycle for the latent scientific teams. The derived results can help users to discover the major research trends for specific fields. And the tracked statuses of the corresponding scientific groups are illustrated in different lifecycle stages, which is useful for the search of research trends or finding of collaboration opportunity according to researchers' different requirements.

Considering the inherent properties of privacy protection and battery efficiency, Device-to-Device (D2D) based proximity service (ProSe) has recently witnessed great development, which enable user to continuously and passively search for and utilize relevant value in one's physical proximity, and is capable to create numerous new mobile service opportunities. However, most of existing ProSe enabled frameworks and applications only provide single-hop Peer-to-Peer connection, lack of easily supporting simultaneous and multi-hop connection on commercially available smartphones to broaden ProSe coverage in static scenario. To solve this issue, this paper proposes a multi-hop connectivity framework BWMesh, through combining Bluetooth and WiFi Direct technologies. BWMesh possesses two special features: exploit users' heterogeneous network and enable smooth interaction among devices with different wireless technologies; provide an easy-to-deploy multi-hop networking framework on currently commercial available smartphones. Specifically, we identify several key issues in multi-hop connectivity framework, design and implement corresponding components to deal with those issues. Then, a prototype MultiChat, is developed based on BWMesh framework, which enables real-time chat among users in proximity, in multi-hop way, without accessing to Internet.

Crowdsourcing is the generalized act of outsourcing tasks, traditionally performed by employees or contractors, to a large group of Internet population (the wise crowd) by means of an open call. With the great development of smartphones with rich built-in sensors and ratio interfaces, mixing smartphone-based mobile technologies and crowdsourcing offers vast computing resources, and leads to a new paradigm called Mobile Crowdsourcing (MCS). In this paper, we presents a taxonomy for the MCS applications, which are explicitly divided as using human as sensors, and exploiting the wisdom of crowd, and typical applications are summarized and compared. Then, a generic MCS framework is proposed, which consists of multiple functional modules. Finally, we elaborate the challenges in MCS applications (i.e., quality control, task management, incentives, as well as security and privacy), and summarize potential solutions. Our main contribution is to classify and provide a summary MCS framework, challenges and possible solutions to facilitate to develop and deploy various MCS applications.

In this paper, an approach to presenting the learning resources, especially those existing user-generated contents associated with learners' activities, as the recommendation to satisfy their current requirements in a social media enhanced learning system, is proposed. Users' attentions are caught and analyzed from the browsing behaviors of learners on a webpage through an eye-tracking device.

Data intensive computing (DIC) offers an attractive option for business to remotely execute applications and load the computing resources from cloud in a streaming way. A key challenge in such environment is to increase the utilization of cloud cluster for the high throughput processing. One way of achieving this goal is to optimize the execution of computing jobs on the cluster. We observe that the order in which these jobs are executed can have a significant impact on their overall completion time (makespan). Our goal is to design a job scheduler that minimizes the makespan. In this study, a new formalization is introduced to present each job as a pair of disk processing and network transmitting two-stage durations. Due to the streaming processing feature, the two-stage operations are executed in an overlap manner and may lead to both one-stage and two-stage scheduling situations. A novel heuristic scheduling strategy is proposed for this hybrid scheduling problem, and the performance of the method is confirmed by the experimental evaluation.

This chapter describes a unified framework for dynamically integrating and meaningfully organizing personal and social Big Data. With the rapid development of emerging computing paradigms, we have been continuously experiencing a change in work, life, playing, and learning in the highly developed information society, which is a kind of seamless integration of the real physical world and cyber digital space. More and more people have been accustomed to sharing their personal contents across the social networks due to the high accessibility of social media along with the increasingly widespread adoption of wireless mobile computing devices. User-generated information has spread more widely and quickly and provided people with opportunities to obtain more knowledge and information than ever before, which leads to an explosive increase of data scale, containing big potential value for individual, business, domestic, and national economy development. Thus, it has become an increasingly important issue to sustainably manage and utilize personal Big Data, in order to mine useful insight and real value to better support information seeking and knowledge discovery. To deal with this situation in the Big Data era, a unified approach to aggregation and integration of personal Big Data from life logs in accordance with individual needs is considered essential and effective, which can benefit the sustainable information sharing and utilization process in the social networking environment. In this chapter, a new concept of organic stream, which is designed as a flexibly extensible data carrier, is introduced and defined to provide a simple but efficient means to formulate, organize, and represent personal Big Data. As an abstract data type, organic streams can be regarded as a logic metaphor, which aims to meaningfully process the raw stream data into an associatively and methodically organized form, but no concrete implementation for physical data structure and stor­age is defined. Under the conceptual model of organic streams, a heuristic method is proposed and applied to extract diversified individual needs from the tremendous amount of social stream data through social media. And an integrated mechanism is developed to aggregate and integrate the relevant data together based on individual needs in a meaningful way, in which personal data can be physically stored and distributed in private personal clouds and logically represented and processed by a set of newly introduced metaphors named heuristic stone, associative drop, and associative ripple the architecture of the system with the foundational modules is described, and the prototype implementation with the experiment’s result is presented to demonstrate the usability and effectiveness of the framework and system.

Comparing to the ordinary text analysis and recommendation, the contents on Social Network Services (SNS) are observably more distinct and less redundant. Content-based recommendation has become the main method on SNSs. Because the limited contents are occurred in SNSs, a considerable effect can’t be reached by using ordinary cluster algorithms. In this paper, we propose a two-phase hybrid clustering algorithm based on Suffix Tree Clustering (STC), which not only uses the words themselves, but relations between them as well. Evaluation experiment and analysis confirm that our techniques have better recommendation results and effects on cold-start scenarios.

In contrast to the huge popularity of Mobile Social Network (MSN) services, not sufficient attention has been reserved to the dynamics of revenue streams and business models in MSN. In this paper, we investigate how a two-sided market (TSM) formulation can be used to provide insight for the problem above from the perspective of MSN platform. Intuitively, MSN service platform could be described as a two-sided market, in which the critical feature is indirect value flow (indirect network externality): one/both sides of the market benefit from increasing adoption and/or consumption of the other side. For instance, participants, like consumers and the third service providers (or advertisers), are the "two sides" of MSN platform. In this paper, our contributions are threefold. First, we thoroughly characterize the economic features of MSN as TSM, including mobility, network externalities and herd effect, as well as long tail property. Second, we provide the various revenue streams within the framework of TSM, including advertising, subscription, and transactions, and characterize the basic components of general business model in MSN. Finally, the free plus premium (Freemium) business model in MSN is formally analyzed. Specifically, we quantitatively characterize the relationship of participation levels among free and premium users as well as service providers, and illustrate the mutual enhancement among those participants. And moreover, Freemium is theoretically compared with the traditional business model with no free users (NF model). The numerical results show that, in Freemium, the participation levels of premium users and service providers, as well as the profit of MSN platform always exceed the corresponding terms in NF model.

As more and more users use the mobile terminals of high computing power, the location-based services (LBS) recommendations for mobile users have become an important and interesting topic. Mobile users are eager to get their interested and reliable services quickly. A considerable number of research works have been dedicated to service recommendation based on users' preferences and locations. In this paper, we study the credibility of recommended services, and propose a set of composite measures on how to provide more reliable services. We further propose the trustworthy Skyline of LBS recommendation in terms of the trust degree based on the newly introduced composite measures to achieve more credibility to provide recommendation services. Experimental results show that our method can recommend desired and trusted services to users.

Recently, mobile social networks (MSN) have gained tremendous attention, which free users from face-to-monitor life, while still can share information and stay in touch with their friends on the go. However most MSN applications regard mobile terminals just as entry points to existing social networks, in which centralized servers (for storage and processing of all application/context data) and continual Internet connectivity are prerequisites for mobile users to exploit MSN services, even though they are within proximity area (like campus, event spot, and community, etc.), and can directly exchange data through various wireless technologies (e.g., Bluetooth, WiFi Direct, etc.). In this paper, we focus on mobile social networking in proximity (MSNP), which is explicitly defined in our paper as: MSNP is wireless peer-to-peer (P2P) network of spontaneously and opportunistically connected nodes, and uses geo-proximity as the primary filter in determining who is discoverable on the social network. In this paper, first, primary support approaches related to MSNP available in literature, are summarized and compared, including MSN, mobile P2P and opportunistic networks. And then, we offer the special characteristics of MSNP, open issues and potential solutions. A networking technologies and platform independent architecture is proposed for developing MSNP applications, and proof-of-concept implementation of WiFi direct based MSNP application is also provided. Our primary goal is to identify the characteristics, technical challenges and potential solutions for future MSNP applications, capable to flexibly adapt to different application domains and deployment requirements.

The 2014 Special Issue of Future Generation Computer Systems is based on 'Hybrid intelligence for growing Internet and its applications'. The first paper highlights the growing needs of healthcare service provision for the elderly under a matured cloud-based environment. It particularly considers the issue in geriatric healthcare in non-adherence to medication regimens especially among those elderly patients who live alone. Another paper discusses the issues of scalable computing and storage capability in a cloud-empowered Internet environment. Integrating QoS awareness towards better requirement of user application development and interference reduction of virtual machine in a cloud-based system is particularly concentrated. Another paper places the emphasis on mining and annotating a semantic relation between data entities on the Internet with temporal, concise, and structured information.

In the Internet of Things (IoT) environment, real-time servers run with many uncertain factors. The modeling and validation for the dynamic service composition of location based services (LBS) in the IoT environment are essential, but it is difficult to consider these uncertain factors. In this paper, after analyzing the advantage and shortage of traditional modeling methods, with the introduction of probability, we propose a Color Probability-TCPN (CP-TCPN) by using the tokens with specific colors. The analysis and functional validation methods based on CP-TCPN are proposed and described. After using it to model and analyze the LBS based in the IoT environment, we demonstrate the result which indicates the CP-TCPN based approach can satisfy the modeling and analysis of the IoT based real-time system with uncertain factors.

Services are expected to be a promising way for people to use information and computing resources in our emerging ubiquitous network society and cloud computing environments. In this study, we propose a metaphoric concept called a flowable service. It is defined as a logical stream that organizes and provides circumjacent services in such a way that they are perceived by individuals to be naturally embedded in their surrounding environments. We present our view on how some of these problems, such as the flexibility, portability and interoperability of services, can be solved using flowable services in order to provide a seamless integration of diverse services in the most intuitive "flowable" way possible, thus achieving maximum satisfaction for both service providers and consumers while decreasing the delivery cost of the services. Recognizing the importance of the awareness of each individual's context for a smooth and accurate provision of services, we further propose a human-centric framework for context-aware flowable services, which harnesses the users' contexts to enable a pro-active support of human activities in a flexible and natural way. Owing to context-awareness, this framework can find and integrate circumjacent services that are considered needed by each individual and can create an ambient service environment that is tailored to his/her specific needs, proclivities and characteristics. (C) 2012 Elsevier Inc. All rights reserved.

Cyber-Individual (Cyber-I) is the digital counterpart of an individual in the real world, which aims at systematically studying and developing comprehensive individual human modeling and its associated applications. The ultimate goal of this research is to create a digital clone for each individual and to provide active desirable services. We present a part of our research work focusing on examining the basic system architecture and the birth process of Cyber-I from a security perspective. In this study, a customized honeypot is used to record multidimensional data Cyber-I is constructed for a corresponding invader. Further, assembling a Cyber-I with associated CI-Applications enables aninvader having more behaviors in the honeypot and provides a possible chance to prolong activities of the invader, which complements a loop mechanism to feed Cyber-I for its growth. The preliminary result in this paper reveals that appropriate authorization and controls are extremely necessary to prevent Cyber-I from being maliciously used and to ensure privacy of its real individual in building an open Cyber-I platform.

Almost all of the existing Collaborative Filtering (CF) methods rely only on the rating data while ignoring some important implicit information in non-rating properties for users and items, which have a significant impact on the preference. In this study, considering that the average rating of users and items have a certain stability, we firstly propose a personalized fitting pattern to predict missing ratings based on the trusty score set, which combines both the user-based CF and item-based CF. In order to further reduce the prediction error, we use the non-rating attributes, such as a user's age, gender and occupation, and an item's release date and price. Moreover, we present the deviation adjustment method based on the Support Vector Regression (SVR). Experiment results show that our proposed algorithms can increase the accuracy of recommendation versus the traditional CF.

With the popularity of Social Network Service (SNS) and the increasing of users, individuals' social roles in a social network have become more and more important in recommendation of the personalized services, and in a collective decision-making process as well. In an SNS system, active users may not represent the major opinions among the whole users, and most of the users' opinions may be multifarious. In this study, we consider users' social roles as an important element to support the collective decision-making process. After introducing the social choice theories and collective decision-making model, we present a three-layer model to analyze users' social roles in a hierarchical way, and utilize it to support the collective decision-making process. A case study for COD (Course-Offering Determination) with an application scenario is demonstrated to show the process of how users' social roles are utilized to support the collective decision-making.

Nowadays, most popular social networking services adopt centralized architecture, in which continual Internet connectivity is prerequisite for each user to exploit those services, and centralized servers are used for storage and processing of all application/context data, even though mobile users are within proximity area (like campus, event spot, and community), and can directly exchange media through various wireless technologies (e.g., Bluetooth, Wi-Fi Direct, etc.). On one hand, the omniscient centralized server may cause serious privacy concern, due to the fact that it collects and stores all users' data (messages, profiles, location, relations, etc.); On the other hand, transmitting a large amount of media generated by users in proximity through Internet, would not only bring a lot of pressure to the network infrastructure and service provider, but incur heavy data traffic cost to users. In this paper, our contributions are twofold. First, we propose a Wi-Fi Direct based P2P social networking framework, which enables direct data exchange among users without using infrastructure network when users are located in proximity, and provide solutions to two core problems in this framework, i.e., discoverability and privacy. Second, the prototype of this framework is preliminarily implemented in Android, which is composed of the following functions: localization based on Google Geocoding, Chat, and file-sharing component supporting intermittent transmission.

The progressive development of information and communication technologies has led us to a new world called hyper world that is composed by the cyber world and the physical world, which, at the same time, brings us the digital explosions of data and connectivity as well as all kinds of smart services. Based on these, Cyber-I was proposed, which aims at creating a unique, digital, comprehensive description for every real person so that the possibility for the person being in the cyber world to get lost in the digital explosions will be reduced. However, the ultimate goal of realizing such Cyber-I needs an endless research process, which will cover multiple areas and disciplines. In order to merge efforts on the study of Cyber-I together as well as provide people with better services utilizing Cyber-I, this paper presents our research and development on a Cyber-I oriented management system where (1) the life cycle of Cyber-I including its birth, growth and death is simulated, (2) with the NO-SQL properties supported by Mongo DB, heterogeneous personal data coming from disparate sources with different formats through varied media can be managed in a scalable way, and (3) apps connected with the Cyber-I can not only fetch personal data but also provide personalized services, while the collected personal data can be used to generate user models from different aspects.

Proximity awareness, the ability to actively (or passively) and continuously search for relevant value in one's physical proximity, is at the core of mobile revolution that is changing the way we interact with people and things around us. Especially, Device-to-Device (D2D) (peer-to-peer (P2P)) solutions support infrastructure-free and self-organized proximity services, and have great commercial potential from application viewpoint. However, astonishingly, the glaring absence of the practical and easy-to-use proximity service development framework is alarming. This paper aims at investigating the Qualcomm AllJoyn middleware, an open source peer-to-peer software development framework for ad-hoc proximity based D2D communication, and exploring how this middleware can enable the development of distributed application in mobile social networks in proximity (MSNP). Specifically, we thoroughly overview the core concepts and basic components in AllJoyn framework, and summarize the typical workflow of developing AllJoyn based proximity application. Furthermore, an AllJoyn based MSNP prototype, AllChat, is designed and implemented, in which users can enjoy text chat and photo sharing with both group and peer interaction modes in real time.

In this study, we propose an integrated approach to organize and refine the social streams in accordance with the analysis of both social data and social roles within a social group. The social stream data will be first collected and organized based on users' individual needs. After the calculation and analysis of users' importance based on the identification and classification of users' dynamical social roles within a user group, a mechanism is developed to organize and refine social streams and further provide users with more suitable and useful information that best fits their needs from more related users, in order to better assist the participatory information search and recommendation process in the social networking environment.

A layout engine combines markup language (HTML, etc.) with formatting information (CSS, etc.), and displays the combined contents on the screen. It is commonly used for many existing web development applications and web browsers. Especially, web development applications use layout engine to confirm the impact on the layout of web editing results to achieve what you see is what you get (WYSIWYG). At present, different layout engines are adopted by various web development applications and web browsers. As there is not a uniform algorithm adopted by layout engine, the results displayed by web development applications are not necessarily consistent with the display on the web browsers. In addition, the results displayed by web development applications depend on layout engines. Hence, it is difficult to edit the common web contents since it cannot objectively reflect the result on web browsers. In this paper, we propose a progressive WYSIWYG solution, by completely separating web layout design from web data editing to generate cross-browser web data. Our proposed approach does not use layout engine, and thus can avoid the disadvantages mentioned above.© 2014 Inderscience Enterprises Ltd.

Location Based Services (LBS) is a kind of real-time service with uncertain factors, and its modeling and validation is essential. In this paper, with the introduction of probability, we propose a Color Probability-TCPN (CP-TCPN) by using the tokens with specific colors as the research objects and redefining several relative parameters. We use CP-TCPN to realize modeling and functional verification of the dynamic services composition for LBS. Simulation result is presented to illustrate the application of CP-TCPN in the modeling and analyzing of the real-time system with uncertain factors. © Springer-Verlag Berlin Heidelberg 2014.

In real human society, influence on each other is an important factor in a variety of social activities. It is obviously important for recommendation. However, the influence factor is rarely taken into account in traditional recommendation algorithms. In this study, we propose an integrated approach for recommendation by analyzing and mining social data and introducing a set of new measures for user influence and social trust. Our experimental results show that our proposed approach outperforms traditional recommendation in terms of accuracy and stability. © 2014 Springer-Verlag Berlin Heidelberg.

In recent years, eye tracking has been widely applied in a variety of fields, such as web usability studies and psychological experiments. To develop a personalized system or network service, it is important to recognize and capture users' needs, situations and contexts in order to create an effective user model. In this paper, we present an integrated approach on how to capture users' unselfconscious information seeking behavior by analyzing their gaze patterns using an eye tracker. We describe the design of an eye tracking experiment, and analyze the eye tracking data to extract gaze patterns, which can be used for use modeling. We further discuss the experiment result and highlight our future work. © 2013 IEEE.

In this study, learning support in a blended learning environment integrated with social media is concentrated on. A method is proposed to integrate the S-UPS (Socialized Ubiquitous Personal Study), which is utilized to collect and manage social stream data, into a blended learning system so as to combine the best aspects of both face-to-face and online instructions for the improvement of learning efficiency and enrichment of learning experience. Based on these, an empirical study is conducted to show the feasibility and effectiveness of our proposed method. Experimental evaluation analysis of the system reveals the enhancement of learning process in the blended learning environment. © 2014 Springer-Verlag.

In this study, we focus on the deep analysis of the learning behavior patterns in the task-oriented learning process, which aims to extract and describe the sharable learning processes for adaptive learning support. The LA-Patterns are extracted to represent an individual's learning behavior patterns. Three categories, named Regular Patterns, Successive Patterns, and Frequent Patterns, are classified to describe users' learning patterns with different features, which can be utilized to recommend users with the adaptive learning process as the learning guidance. The experiment and analysis results in a learning management system are discussed finally. © 2014 Springer International Publishing.

Differing from many studies of recommendation that provided the final results directly, our study focuses on providing an optimized process of information seeking to users. Based on process mining, we propose an integrated adaptive framework to support and facilitate individualized recommendation based on the gradual adaptation model that gradually adapts to a target user's transition of needs and behaviors of information access, including various search-related activities, over different time spans. In detail, successful information seeking processes are extracted from the information seeking histories of users. Furthermore, these successful information seeking processes are optimized as a series of action units to support the target users whose information access behavior patterns are similar to the reference users. Based on these, the optimized information seeking processes are navigated to the target users according to their transitions of interest focus. In addition to describing some definitions and measures introduced, we go further to present an optimized process recommendation model and show the system architecture. Finally, we discuss the simulation and scenario for the proposed system.

Researchers tend to agree that an increasing quantity of data has caused the complexity and difficulty for information discovery, management, and reuse. An essential factor relates to the increasing channels (i.e., Internet, social media, etc.) for information sharing. Finding information, especially those meaningful or useful one, that meets ultimate goal (or task) of user becomes harder then it is used to be. In this research, issues concerning the use of user-generated contents for individual search support are investigated. In order to make efficient use of user-generated contents, an intelligent state machine, as a hybridization of graph model (Document Graph) and petri-net model (Document Sensitive Petri-Net), is proposed. It is utilized to clarify the vague usage scenario between user-generated contents, such as discussions, posts, etc., and to identify correlations and experiences within them. As a practical contribution, an interactive search algorithm that generates potential solutions for individual is implemented. The feasibility of this research is demonstrated by a series of experiments and empirical studies with around 350,000 user-generated contents (i.e., documents) collected from the Internet and 200 users.

"U-Business" is a novel type business environment, which can provide various services via many mobile devices. In order to provide personalized service to different users, user model (UM) can play an important role in U-Business. UM reflects some characteristics of users to a certain degree, which is used widely in U-Business, like personalized recommendation, social computing, information retrieval services, and so on. Currently, there are more and more researchers who focus on the building and update of UM based on the activities of people. However, as too many UM appeared, the number of UM in cyber space is increasingly large, which takes a lot of space and cost. Furthermore, after some users disappear in the physic world, their models are still working in the cyber world. This case is not reasonable obviously, but few researches take care about it. Therefore, one of important issues, the death of UM should be taken into account in the whole life-cycle of user model. This paper proposes a specific user modeling method for the Cyber Individuals (Cyber-I) in U-Business. The essential difference between this UM and traditional ones is that it has a life, that is, birth, growth, and demise, like a life-cycle of Cyber-I. Specially, the significance of UM life ending and five states of UM death are described from an organic viewpoint. In addition, there is a framework of the whole life process of UM. Finally, the proposed idea is applied to the field of personalized service.

Sharing information and resources on the Internet has become an important activity for education. The use of ubiquitous devices makes it possible for learning participants to be engaged in an open and connected social environment, and also allows the learning activities to be performed at any time and any place. In this study, the discovery of correlation among shared resources is concentrated. A hypothetical scenario is considered that the information, such as photos and thoughts, is applicable to be shared with implicit context (i.e., geographical information) by learners through a practical implementation, PadSCORM, on a mobile device. Two major contributions are achieved. First, the correlations among resources are determined through usage experiences mining and geographical information adjustment. It then assists learners in filtering out redundant information by highlighting the significance of resources. Second, an intelligent searching algorithm is proposed to visualize adaptive routes to facilitate search process and to enrich the learning activity. The empirical experiments revealing the feasibility and performance (e.g., accuracy and effectiveness) are conducted in the universities in North Taiwan. © 2012 John Wiley &amp
Sons, Ltd.

Recently, social networking sites such as Facebook and Twitter are becoming increasingly popular. The high accessibility of these sites has allowed the so-called social streams being spread across the Internet more quickly and widely, as more and more of the populations are being engaged into this vortex of the social networking revolution. Information seeking never means simply typing a few keywords into a search engine in this stream world. In this study, we try to find a way to utilize these diversified social streams to assist the search process without relying solely on the inputted keywords. We propose a method to analyze and extract meaningful information in accordance with users' current needs and interests from social streams using two developed algorithms, and go further to integrate these organized stream data which are described as associative ripples into the search system, in order to improve the relevance of the results obtained from the search engine and feedback users with a new perspective of the sought issues to guide the further information seeking process, which can benefit both search experience enrichment and search process facilitation. © 2012 Springer Science+Business Media, LLC.

In this study, in order to support and facilitate the web-based learning, we concentrate on user learning behavior pattern discovery in a task-oriented learning process. Based on a hierarchical graph model which can describe relations among learning actions, learning activities, learning sub-tasks and learning tasks, we introduce the formal definitions for Learning Action Pattern and Goal-driven Learning Group to discover and represent users' learning behavior patterns within a learning task process. Two integrated algorithms are developed to calculate and generate the Learning Action Patterns for an individual user and the Goal-driven Learning Groups for a number of users, which can benefit sharing of learning activities and improve learning efficiency in e-learning environments. Finally, the design of a prototype system with experiment results is discussed. © 2013 Springer-Verlag.

Recently, eye-tracking has been widely applied in a wide spectrum of fields for both academic researches and Business. In this study, we concentrate on the analysis of instant (and often subconscious) information generated from interactions between an individual and devices, such as a PC, laptop, and mobile phone. We present the experiment design to capture and extract the viewing patterns in Twitter using the eye-tracking technology. We show a set of experiment results based on the analysis of eye gazing data, in order to demonstrate how the subjects look for specified keywords in the Twitter timeline, which can further contribute to categorization of viewing patterns.

With the high accessibility of the social media, more and more people have been accustomed to sharing their personal contents across the social networks, which results in an explosive increase of data scale. In this study, in order to support information and knowledge discovery in big data, we propose an approach to aggregation and integration of personal big data from life logs in accordance with individual needs, which can benefit the sustainable information utilization process. In details, the organic stream, which is designed as an extensible data carrier, is introduced and developed to formulize and organize the personal big data, in order to extract dynamical individual needs from the tremendous amount of data posted through social media, and further aggregate and integrate the related data in a meaningful way, which can also facilitate the personalized information retrieval and reuse process. The architecture of the system with the foundational modules is given, and the experiment result is presented to demonstrate the usability and effectiveness of our approach.

With the rapid growing of users in the SNS environments, their network relationships in Internet have become as complex as in the real world. On the other hand, the user's social role plays an important role in providing individualized services, such as information recommendation in the SNS environments. However, as we know, people will switch or change their social roles dynamically in the real society, which is also true in the cyber network space, such as SNS. In this paper, we present a basic model to describe social roles in both the real world and the SNS environment, in which a set of attributes and factors are considered and introduced to represent the time-changing social roles in different situations and contexts. The mechanism that we develop to identify the roles is twofold: mapping of real world situations to cyber space, and analyzing social role attributes based on the mapping and synthesizing. We further describe an application scenario in regard to how to map the different situations and analyze the role attributes, in order to dynamically identify roles in the cyber space.

Recently, mobile social networking applications have gained tremendous attention, which free users from face-to-monitor life, while they still can share information and stay in touch with their friends on the go. However most MSN applications regard mobile terminals just as entry points to existing social networks, depend on centralized servers (for storage and processing of all application/context data) and continual Internet connectivity, even though mobile users are within proximity area (like campus, event spot, and community), and can directly exchange data through various wireless technologies (e.g., WiFi Direct, etc.). In this paper, we focus on thoroughly overviewing the mobile social networking in proximity (MSNP), including existing applications, characteristics and challenges. Our primary goal is to identify the characteristics, technical challenges and potential solution for future MSNP applications, capable to flexibly adapt to different application domains and deployment requirements.

Usually, in mobile server cloud computing (MSCC) environments, there exist enormous sensors conducting various tasks. Localization of sensor nodes using the technologies in Wireless Sensor Network (WSN) is critical to both cloud infrastructure operations and most applications. Specifically, in WSN, Malguki (the method's name, Malguki, means 'spring') is an effective range-based algorithm which can compute the location of a node using noisy distance estimations. However, through simulation, we found that, in original Malguki algorithm that uses an iterative process to locate unknown nodes, the initial positions of unknown nodes in iteration are evenly and randomly selected, which may cause large average localization error. Considering the mentioned weak point, this paper proposes to enhance the Malguki with a simple range-free Centroid localization algorithm, which intentionally obtains initial positions of unknown nodes in iteration by Centroid algorithm. Simulation results show that the integration of range-based and range-free localization algorithm, Centroid Malguki always performs better than original Malguki algorithm. © 2013 IEEE.

Background: In the cloud computing environments, numerous ambient services may be created speedily and provided to a variety of users. In such a situation, people may be annoyed by how to make a proper and optimal selection quickly and economically. Methods: In this study, we propose an Adaptively Emerging Mechanism (AEM) to reduce this selection burden with an interdisciplinary approach. AEM is applied and integrated into the Flowable Service Model (FSM), which has been proposed and developed in our previous study. We consider the user’s feedback information is a pivotal factor for AEM, which contains the user’s satisfaction degree after using the services. At the same time, we assume that these factors, such as the service cost, matching result precision, responding time, personal and social context information, etc., are essential parts of the optimizing process for the selection of ambient services. Results and Conclusion: By analyzing the result of AEM simulation, we reveal that AEM can (1) substantially improve the selection process for LOW feedback users
(2) bring no negative effect on the selection process for MEDIUM or HIGH feedback users
and (3) enhance the rationality for services selection.

Recently, SNS (Social Network Service), blog and microblog have become very popular. Stream data, a large collection of diverse contents that are created dynamically in the form of streams, have become an important part of the Internet resources. At the same time, it has become easier to collect people's activities as their lifelogs, not only in the cyber space, but also in the physical world by means of ubiquitous and sensing technology. Either stream data or lifelogs represent different aspects of people's information behaviors and social activities, which we call Social Streams. In this study, we try to integrate and organize these social stream data, such as Twitter Tweets, into Ubiquitous Personal Study (UPS) proposed in our previous study. In this paper, we introduce and define a set of new metaphors: Drop, Stream, River and Ocean, to represent a variety of social stream data in different stages, in order to enable UPS socialized toward an individualized information portal. We further propose a Framework of Organic Streams to meaningfully organize these stream data. We discuss the design and implementation issues of a prototype system, and describe the algorithms to realize our proposed metaphors. Moreover, we show a scenario of using the socialized UPS to support learning activities, with experimental data and analysis results. (C) 2011 Elsevier Inc. All rights reserved.

In the study, we propose an integrated adaptive framework to support and facilitate individualized learning through sharing the successful process of learning activities based on similar learning patterns in the ubiquitous learning environments empowered by Internet of Things (IoT). This framework is based on a dynamic Bayesian network that gradually adapts to a target student's needs and information access behaviours. By analysing the log data of learning activities and extracting students' learning patterns, our analysis results show that most of students often use their preferred learning patterns in their learning activities, and the learning achievement is affected by the learning process. Based on these findings, we try to optimise the process of learning activities using the extracted learning patterns, infer the learning goal of target students, and provide a goal-driven navigation of individualized learning process according to the similarity of the extracted learning patterns.

Recently, P2P (Peer-to-Peer) technology has witnessed a rapid development. Basically, one of key components in successful P2P applications is how to efficiently look up resources. Considering that structured P2P is a relatively efficient way to locate resources, this paper conducted two improvements to increase the search efficiency in Chord-based algorithms, one of the most popular structured P2P resource lookup protocols. In detail, our contributions are twofold. First, considering the fact that routing information in Chord is not abundant enough for efficient resource search, and looking up resource can only be enforced in clockwise direction, a new algorithm called AB-Chord is proposed to reconstruct the finger tables in Chord, in which counter-clockwise finger table is added to achieve resource queries in both directions, and the density of neighboring fingers is increased. Additionally, AB-Chord implements a new operation to remove the redundant fingers introduced by adding fingers in AB-Chord. Experimental results show that AB-Chord's query efficiency has been improved in terms of the average lookup hops and average lookup delay. And furthermore, considering that the proposed AB-Chord algorithm enlarged the finger table which may cause the forwarding-storm of routing maintenance messages, we further propose AB-Chord+, which appropriately extends the periodic time of updating finger tables and makes the joining and leaving nodes actively send updating messages, to reduce the number of messages forwarded in the network. Simulated results show that AB-Chord+ reduced the network bandwidth consumption.

There are various factors that may affect user experience (UX) on a product. In this study, we investigate the attachment factor in UX over a time span. In this paper, we firstly discuss our view on attachment and UX from an interdisciplinary perspective, by taking iPhone as our case study. We describe the design of an experiment based on a time span sheet and take account into account the attachment factor in UX. We further reports the experiment results based on data collected from 36 subjects (among them 20 are iPhone users), and discuss our observations from the statistical analysis. The experiment result shows that there exists interrelation between attachment factor and UX, and the attachment of users to a specific product is created and strengthened while using it for a long time. Copyright 2012 ACM.

Conference proceedings front matter may contain various advertisements, welcome messages, committee or program information, and other miscellaneous conference information. This may in some cases also include the cover art, table of contents, copyright statements, title-page or half title-pages, blank pages, venue maps or other general information relating to the conference that was part of the original conference proceedings. © 2012 IEEE.

Recommender system is a focus in the age of information explosion. In this study, with the benefit of social networking service, we propose a User-Centric Integrated Recommendation Model based on combining of users' individualities and commonalities, in which users' interests are focused and their transitions are traced by analyzing users' information access behaviors and histories, and then a sequence of information seeking actions are recommended to target users through dectecting the transitions of their interests focus by interaction of users and the system, and extracting successful experience from a reference user group, in which the reference users are similar to the target users. A set of bookmark tags are used to describe relations of Web pages. The pages accessed by users are classified by the bookmark tags, and grouped into two categories of individual and common interests and their sub-categories. The individual interests are divided into three types: strong interest, weak interest and uncertain interest. The common interests are divided into popular interest, public interest and private interest. In this paper, in addition to describing definitions and measures, we present a mechanism of inferring interest focus and show the system architecture. Finally, the conclusion and further work are introduced. © 2012 IEEE.

In open networking environments, resources, services, strategies and actions are voluntarily provided, maintained, chosen and determined by independent, rational and autonomous peers. Basically, Game theory is a basic tool for modeling choices by rational agents in those environments, which, usually assumes players choose strategies which maximize utility of game outcomes given their beliefs about what others players will do. This means that the most challenging question is often how beliefs are formed. Intuitively, beliefs depend not only on what people know to be true, but also on what they want to be true (desire). This paper introduces a model of rational choices that allows for this possibility that peers' beliefs are affected by their interests (desires), and, based on this model, analyzes the impact of psychological belief (optimistic bias and pessimistic bias) on the existing altruism-based cooperative mechanisms in Peer-to-Peer (P2P) systems. We found that, when there exist free riders, on the contrary to the intuitive thought, pessimistic bias of belief on contribution level can facilitate the P2P systems to converge to the stable equilibrium in easier way than optimistic bias. © 2012 IEEE.

Comparing with traditional e-learning, the socialized e-learning system provides us a new paradigm to deliver knowledge through social medias. In this study, we concentrate on the organization of knowledge delivered across the social media enhanced learning environment in order to support the learning process. An integrated approach is proposed to capture users' intentions using eye-tracking technology, and further organize the related social streams into meaningful contents, in which both re-defined and newly created knowledge from interactions among users can be extracted in accordance with users' individualized requirements to assist their learning tasks and finally promote the learning efficiency for both professors and students. © 2012 IEEE.

In this study, we try to discover the potential and dynamical user correlations using those reorganized social streams in accordance with users' current interests and needs, in order to assist the information seeking process. We develop a mechanism to build a Dynamical Socialized User Networking (DSUN) model, and define a set of measures (such as interest degree, and popularity degree) and concepts (such as complementary tie, weak tie, and strong tie), which can discover and represent users' current profiling and dynamical correlations. The corresponding algorithms are developed respectively. Based on these, we finally discuss an application scenario of the DSUN model with experiment results. © 2012 Springer-Verlag.

In this study, we concentrate on learning activity sharing and individualized recommendation based on dynamical user correlations, in order to support and facilitate the web-based learning process integrated with social streams. A user correlation-based learning activity model is built to demonstrate the relations among user, learning task and learning activity. Based on these, an integrated method is proposed to provide a target user with the possible learning activity as the next learning step, which is expected to enhance the learning efficiency. Finally, design of a Moodle-based prototype system is discussed. © 2012 Springer-Verlag.

In this paper we argue a user has to be in the center of information seeking task, as in any other task where the user is involved. In addition, an essential part of user-centrism is considering a user not only in his/her individual scope, but expanding it to the user's community participation quintessence. Through our research we make an endeavor to develop a holistic approach from how to harnesses relevance feedback from users in order to estimate their interests, construct user profiles reflecting those interests to applying them for information acquisition in online collaborative information seeking context. Here we discuss a human-centric integrated approach for Web information search and sharing incorporating the important user-centric elements, namely a user's individual context and 'social' factor realized with collaborative contributions and co-evaluations, into Web information search.

Learners can benefit more from the emerging ubiquitous computing technology in more learning scenarios beyond traditional computer-based e-learning system. But due to the great diversity of device specification, learning contents, and mobile context existing today, learners may have a poor learning experience in the ubiquitous computing technology-enhanced learning (u-learning) environment. This paper proposes a multiple response approach for adaptive learning service with context-based multimedia contents, which recommends preferred media for learners according to the u-learning environment. Based on six learning statuses from SCORM, five learning responses of learning objects are used to reward or penalize the preferred media according to the learning context. In the evaluation experiment, the accessed object, time, location and mobile device are mainly used as context data. With the comparison between the controlled group and non-controlled group, the results show that the proposed method can improve the utilization rate of learning objects which implies that the learners are more interested in these recommended media. The results also show that the learning experience is improved. © 2011 Springer Science+Business Media B.V.

Identity fraud has become one of major concerns for broad communities. The new information era needs a new digital identity infrastructure to support next generation Internet. This article suggests a hierarchical structure for digital identities. We define and classify all digital identities into three broad categories: Object, People and Organization. This paper is the first to systematically address the classification of digital identities. More and more individuals and communities heavily rely on the network. Many countries are trying their own digital identity initiatives, like E-passport, national smart card, etc. This article intends to initiate a discussion on a universal digital identity infrastructure for our future. We believe that in the near future all paper-based identities will be replaced by digital identities. A robust digital identity infrastructure will take a vital role in the future information age. © 2011 IEEE.

With the continued increase in the use of ubiquitous devices, learners' expectations of learning services have also increased. Most of the learning services being provided today are not designed or delivered for ubiquitous learning environments because these learning services have not been taken into account rich learning contexts. This paper proposes a service-oriented architecture for context-aware adaptive learning services, which contains adaptation service, transforming service, and presentation service. Within each service, the conflicts are resolved by different learning rule base. The results show that the system may improve the learning experience under ubiquitous learning environments, and the learners are more interested in adapted learning service than those from general learning service. © 2011 IEEE.

There have been two open questions about Web 2.0 enhanced CoP (Community of Practice). The first one is the isolation of CoPs caused by the variety of web services, and the second one is the context loss in knowledge transference. Both of the questions arise from the discontinuity of the knowledge transference in a CoP or among CoPs. In order to tackle these questions, CAPK (Context, Actor, Pipe and Knowledge object), a new learning process model, is proposed to model learning in the Web 2.0 technology enhanced CoPs. In this model, the premise of learning is the connection to the knowledge objects by various pipes which can be used to maintain a connection with a knowledge object. Context is considered as the most important component of effective learning in this model. CAPK model can provide us a new perspective to tackle the above problems. In order to achieve effective learning in current web based CoPs, flexible connection to and management of knowledge objects are necessary to support learning in CoPs. Besides, a uniform context description is also a requirement. In this paper, after the introduction of the CAPK learning process model, we describe design and implementation of a prototype system for managing the pipes of knowledge objects to provide a solution to transfer context information through Activity Streams specification in and cross various systems. © 2011 IEEE.

Social learning is showing one of the immersive potentialities for human knowledge gathering on the Internet throughout the last decEDe. Due to rapid growth of Internet, people can communicate with each other using IM (Instant Messaging), WWW and P2P. IM Services, such as
Yahoo, Twitter, Google Talk, ICQ (I seek you), etc. which provides individual and group communications. Web search engines (Google, Yahoo, AOL etc.) collect versatile information by crawling web directories, and web public documents (html, PDF, text, etc.) throughout the world. P2P (Gnutella, Napster, Freenet, etc.) serves file sharing between and/or among users. However, IM, WEB and P2P do not support WW, which indicates WHO the right person/user is and WHAT the right topic a user is looking for. To overcome the WW issues, this research presents a hybrid search engine called P2P-SSE (P2P Social Search Engine) which can find the WW and serves users queries for the support of social learning. © 2011 IEEE.

This paper describes an integrated approach to provide an optimized learning process to students by analyzing the log data of learning activities and extracting their learning patterns. Our analysis results show that most of students almost always use their main learning patterns in their learning activities, and the learning achievement is affected by the learning process. Based on these findings, we try to optimize the process of learning actions using the extracted learning patterns, infer the learning goal of students, and then navigate them a personalized learning process according to the similarity of the extracted learning patterns. © 2011 IEEE.

Sharing learning activity in and cross various systems is a prospective approach to implement OLP (Open Learning Process), a paradigm to share the knowledge generated in the learning process. The question is that the systems supporting teaching and learning are always separated and isolated. To implement OLP, firstly a uniform description of learning activity and a protocol to share learning activity in and cross systems are necessary. Secondly, a systematic framework to support learning activity description is also important. In this research, first, we show our vision of OLP by introducing metaphors such as a drop or an ocean of learning activity. Then based on the analysis of web based learning activity through Activity Theory, we propose a learning activity model to describe learning activity. Moreover, a specification of learning activity streams based on Activity Streams specification is introduced. Finally, a prototype system supporting learning activity streams is implemented based on UPS (Ubiquitous Personal Study) platform. © 2011 Springer-Verlag.

Sharing resources and information on Internet has become an important activity for education. The MINE Registry, a branch of distributed repository, inherited the architecture of CORDRA has been developed for storing and sharing Learning Objects. Following the usage experiences, especially those being utilized to generate the lecture, the interaction structure is defined to clarify the relationships among Learning Objects. The methods to social network analysis are applied to quantify the implicit correlations and to evaluate the interdependency. In addition, an intelligent mining algorithm is proposed to explore the developed interaction structure and automatically generates lecture templates corresponding to the query criteria. The concentration of this study is to facilitate the complex and time-consuming process of creating lectures through a simple search mechanism. The implemented system has demonstrated the preliminary results and the feasibility are also revealed by the evaluation results. © 2011 Springer-Verlag.

Sharing resources and information on Internet has become an important activity for education. The learning object repository has been developed to achieve efficient management of learning objects. Following usage experiences of learning objects collected in the past, this study concentrates on investigating implicit information between learning objects. We define a social structure for identifying relationship between learning objects and define a set of metrics to evaluate the interdependency. The structure identifies usage experiences and can be graphed in terms of implicit and explicit relations among learning objects. As a practical contribution, an adaptive algorithm is proposed to mine the social structure. The algorithm generates adaptive learning sequence by identifying possible interactive search input and assists them in completing self-paced learning situation. © 2011 IEEE.

Sharing information and resources on the Internet has become an important activity for education. The use of ubiquitous devices makes possible for learning participants to be engaged in an open and connected social environment, and also allows the learning activities to be performed at any time and any places. In this study, the discovery of correlation among shared resources is concentrated. A hypothetical scenario is considered that the information, such as photos and thoughts, is applicable to be shared with implicit context (i.e. geographical information) by learners through a practical implementation, PadSCORM, on a mobile device. Two major contributions are achieved. First, the correlations among resources are determined through usage experiences mining and geographical information adjustment. It then assists learners in filtering out redundant information by highlighting the significance of resources. Second, an intelligent searching algorithm is proposed to visualize adaptive routes in order to facilitate search process and to enrich the learning activity. © 2011 ACM.

Service computing aims to provide IT and computing resources to users in a way that it simply serves them. However, many issues remain in, such as portability, interoperability, and heterogeneity of diverse services, in addition to service modeling, creation, deployment, discovery, recommendation, composition, and delivery, in order to provide a service that best fits user needs and contexts at the time. In this study, we propose a new model of flowable services. Its primary purpose is to realize seamless integration and provision of diverse services in an intuitive "flowable" way maximally close to the "flow" of users' activities and thoughts, and as pertinent as possible to their needs, thus gain the most satisfaction. In this paper, we propose and discuss a human-centric framework of flowable services toward service assurance. We further describe the evaluation of the model, and an application scenario for adaptive delivery of personalized learning service under the proposed framework.

Ubiquitous Learning (U-Learning), as an emerging learning paradigm, makes it possible for learners to carry out the learning activities at any places and at anytime. With the advantages of the devices, learners can obtain a variety of supplementary materials from the Internet. In the scope of distance learning, LOR (Learning Object Repository) stands for managing and sharing of learning related materials (known as learning objects). However, some challenges may raise while performing these activities. For instance, a huge amount of learning objects may appear while learners utilize the search service provided by LOR. Learners have to spend time on collecting relevant resources for specific purposes. This situation may discourage the reusability of learning objects especially in a ubiquitous environment. In this paper, based on systematic re-examination of reuse scenarios, an adaptive mechanism, as a resource discovery and search middleware, was proposed to assist learners in obtaining possible objects under ubiquitous environment. Achievement of the proposed mechanism can produce search results adaptive to specific situations in order of similarity degree based on the mixed information. We try to filter out some irrelevant results by using the past usage history, current geographical information and input query, so as to enhance the efficiency of learning objects retrieval in a ubiquitous environment. As a pilot test, Apple iPhone was utilized to be the major client testbed. © 2011 Academy Publisher.

Recently, the vast dialog in the microblog, such as twitter, Facebook has become increasingly popular. As we post more messages in microblogs, information is spreading more quickly and widely. These widely spread and diversified contents could be viewed as data streams, which have become an important part of the Internet resources. However, these separated data streams are littery and meaningless, so we need to collect and organize them together to provide us with meaningful information. It is hard to imagine that we could find useful information by simply inputting a few keywords into a search engine in such a stream environment. In this study, we try to find a way to seek the information related to users' personal and current interests and needs among these data streams and provide users with other more relevant information. We introduce a set of metaphors to represent a variety of data streams in different levels, and define two new metaphors: heuristic stone and associative ripple to assist the seeking process and describe the results. Based on these, we further propose two algorithms for the information seeking and processing, and discuss a scenario of the information seeking process that utilizes the proposed metaphors and algorithms. Copyright 2011 ACM.

Recently, social media enhanced learning has become more and more popular. It is featured as learning through interaction and collaboration in a community or across a social network, which can be considered as a kind of social learning. In this study, we integrate SNS (such as twitter) into the web-based learning process and further delve into the discovery of potential information from the reorganized stream data. We propose a Dynamical Socialized User Networking (DSUN) model which represents users' profiling and dynamical relationship by a set of measures. Finally, we show an application scenario of the DSUN model to assist the learning process and enhance the learning efficiency in web-based environments. © 2011 Springer-Verlag.

In this study, we integrate social media into the blended learning environment, and further delve into utilizing the eye-tracking technology to enhance learning to be socialized and more efficient. We propose an approach to employ eye-tracking to extract those related learning stream data according to different seeking patterns in a learning process. Based on these, we go further to organize these raw stream data into meaningful learning contents in accordance with specific tasks in order to benefit both teachers and students, which can assist the learning process and promote the learning efficiency in the blended learning environment. © 2011 IEEE.

In this paper, we introduce a session key distribution mechanism for fast secure handover in wireless mobile networks. The proposed mechanism is based on the stream control transmission protocol in where a mobile node actively changes its IP address without its connection loss. When a mobile node moves between different access routers, the required session key at a new access router is distributed previously through the tunnel established between the previous access router and the new access router. Due to the reduced key distribution time, the mobile node achieves its secure seamless handover. The provided performance analysis proves that the proposed mechanism provides the reduced signaling cost compared to existing mobility protocols. In addition, the proposed mechanism reduces the handover latency so that the reduced amount of packet loss provides the stable handover performance for real-time applications.

t is a common sense that drowsiness during driving could cause traffic accidents. Related researches show that drowsiness occurs when the parasympathetic nerve system predominates over others, and the high frequency (HF) component of heart rate variability (HRV) on an electrocardiogram (ECG) is closely related with the parasympathetic nerve activity, so it is reasonable to evaluate drowsiness based on heart rate analysis. In the driving simulation study, ECG and calculated R-R intervals were put on record during the monitoring process of drowsiness degrees. Relations among different indexes, such as, temporal change of HF band, RR50, Coefficient of Variation of R-R interval (CVRR), and LF to HF ratio (LF/HF) were evaluated as well. The experimental results suggest that there exist certain correlations between drowsiness and HF, RR50, CVRR, or LF/HF, which will be compared in the future with data collected in real driving experiments.

This study aims to provide learners with appropriate learning activities navigation based on a learning process recommendation system that gradually adapts to their needs. A learning activity consists of a series of purposeful learning actions with a specific sequence. In the system, user profile is generated from various learning related activities that are collected over different time spans. A dynamic user group is generated based on the similarity of user profiles which have had successful learning experience, and can be used to create a personalized navigation of learning activities for a target learner. In this paper, after introducing how to create a dynamic design of learning activities navigation, we focus on discussing the design and implementation issues of the prototype system for personalized learning support.

In this study, we develop an innovative approach using sharable learning activity to create an open learning network. In this paper, we present our vision of open learning and a framework for sharable learning activities. We further discuss how to utilize open source software to implement a prototype system, and describe the scenarios and features of the proposed framework and system.

Layout engine combines markup language (Html, etc) with formatting information (CSS, etc), and displays the combined contents on the screen. It is commonly used for many existing web development applications and web browsers. Especially, Web Development Applications use Layout Engine to confirm impact on the layout of web editing results to achieve What You See Is What You Get (WYSIWYG). At present, different layout engines are adopted by various web development applications and web browsers. As there is not a uniform algorithm adopted by layout engine, the results displayed by web development applications are not necessarily kept the same as displayed on the web browsers. In addition, the results displayed by web development applications are depended on layout engines. Hence it is difficult to edit the common web contents since it cannot objectively reflect the result on web browsers. In this paper, we proposes a novel WYSIWYG solution, by completely separating web layout design from web data editing to generate cross-browser web data. Our proposed approach does not use layout engine, and thus can avoid the disadvantages mentioned above.

Recently, there have been lots of researches on how to provide personalized services to different users according to their personal characters, needs, situations, contexts and so on. Such personalized services are often based on users' profiles provided by the users in the beginning, and users' records or experiences in using corresponding systems in certain periods. However, different systems or terminals collect user information independently and cannot share the user's personal information collected by these different systems/terminals. As a result, each system can only utilize the limited information collected by the system itself to provide services or recommendations, and it thus cannot meet what users' needs in varied situations across the different systems/terminals. Therefore, it is still an open issue on how to gather, share and utilize various kinds of personal information to effective personalized services in right time, right place and right means to users. In this paper, a case study for personalized pervasive learning is discussed as one typical application of the concept of Cyber-I, an individual's counterpart on the cyberspace. The Cyber-I aims to provide a better environment for users to obtain what they may really need on the Web, and it can be considered as an innovated possibility of web usage scenario in the near future. Our case study using pervasive devices (iPad, cell phone, and laptop) for learning issues can be regarded as the best practice to support the proposed Cyber-I. © 2010 IEEE.

Recently, SNS, blog and microblog have become very popular. A different form of diversified contents, comparing to traditional static HTML web pages, are so easy to be generated dynamically, and can be provided to users by timeline. These contents are generally processed data streams, and have become an important part of the Internet resources. In this study, we try to integrate these dynamically diversified contents, such as twitter stream data, into Ubiquitous Personal Study (UPS), a personalized virtual study proposed in our previous study. In this paper, we introduce and define a set of new metaphors to represent a variety of data streams, such as Drop, Stream, River, and Ocean. We propose a Framework of Organic Streams to meaningfully organize these stream data. We describe the design and implementation issues of a prototype system, and discuss a scenario of using the enhanced UPS to support learning that utilizes twitter data streams. © 2010 IEEE.

With the development of internet and search engine, users are thought to obtain everything through corresponding web services. Although general purpose searching such as one provided by Google is powerful, searching mechanism for different purposes has to rely on specific metadata. We followed SCORM and CORDRA to develop a registry system named MINE Registry for storing and managing the learning objects. As a contribution, we propose the concept of "Reusability Tree" to represent the relationships among relevant Learning Objects and to enhance CORDRA. We further collect relevant information while users are utilizing Learning Objects, such as citations and time period persisted. The feedbacks from community users are also considered as critical elements for evaluating significance degree of Learning Objects. Through these factors, we propose a mechanism to weight and rank Learning Objects in the MINE Registry, in addition to other external learning objects repositories.

Services are expected to be a promising way for people to use information and computing resources in the emerging ubiquitous network society. In this study, we propose a metaphoric concept called flowable service. It is defined as a logical stream that organizes and provides circumjacent services in such a way that they are perceived by individuals as those naturally embedded in their surrounding environments. We present our view on how some of these problems, such as flexibility, portability and interoperability of services, can be solved using the flowable services to make seamless integration of diverse services in an intuitive "flowable" way possible, thus achieving maximum satisfaction of both service providers and consumers while decreasing the delivery cost of services. Recognizing the importance of the awareness of each individual's context for smooth and accurate provision of services, we conceive the flowable service framework that harnesses users' contexts to enable pro-active support of human activities in a flexible and natural way. Owing to context-awareness, the framework is able to find and integrate the circumjacent services that are considered to be needed by each individual, and create an ambient service environment tailored to his/her particular needs, proclivities and characteristics. © 2010 IEEE.

In line with the popularity of Internet and the development of search engine, users request information through Web-based services. Although general purpose searching such as one provided by Google is powerful, searching mechanism for specific purposes could rely on metadata. We followed SCORM and CORDRA specifications to develop a registry system, called the MINE Registry, for storing and sharing 20,738 Learning Objects created in the past five years. As a contribution, we propose the concept of "Reusability Tree" to represent the relationships among relevant Learning Objects and to enhance CORDRA. We further collect relevant information while users are utilizing Learning Objects, such as citations and time period persisted. The feedbacks from community users are also considered as critical elements for evaluating significance degree of Learning Objects. Through these factors, we propose a mechanism to weight and rank Learning Objects in the MINE Registry, in addition to other external learning objects repositories. © 2010 IEEE.

Cloud computing, which can provide ubiquitous services, is a nexus of hardware, software, data and people. Service model is an essential part of cloud computing. It is recognized that the service model should provide precise computing results with low costs. However, we think it is not adequate as a computing paradigm if no consideration is made on the human aspects, especially, the factors of end users. In this study, we propose a novel service model: Flowable Service Model (FSM),which aims to provide appropriate services (as a flow) for users in any time and any place, and let them to enjoy ambient life empowered by cloud computing. How to make the services flowing from the cyber world to the real world? Being inspired by James Williams' stream of consciousness theory, we conceive FSM to leverage a variety of personal information, such as users' behaviors and habits. The operational mechanism for FMS is described in this paper. We expect our proposed model can make every user to feel more ambient with available cloud services. © 2010 IEEE.

This paper proposes an agent based support system for promoting interactive teaching and learning. The agent based system is developed with the emphasis on learning to know students' characters, habits, and preferences so that the teacher or the system can give students appropriate hints, encouragements, confidents, and as well as influence students gradually without their noticing and feeling of be persuaded and pressured. With more active participation of students in the interactive learning process, the teaching is more effective and the learning is more efficient. ©2010 IEEE.

This study presents an integrated adaptive system to support and facilitate individual learning based on a recommendation model that gradually adapts to a learner's needs and information access behaviors, including various learning related activities, over different time spans. In this paper, we focus on describing the integrated framework and discussing the design and implementation issues of the prototype system for personalized learning support. ©2010 IEEE.

The content granularity and context information are two important factors to the efficiency and reusability of learning objects. The context information is necessary to facilitate the discovery and reuse of learning objects stored in global and/or local repositories. However, traditional learning objects are generally not conceived to incorporate with enough context information. Users have to do some extension of the description item set to fit their special use. In this paper, in order to deal with the issue mentioned above, we firstly introduce a context-rich paradigm, the related service driven tagging strategy, and a context model of learning objects. We further explain how to use the context information to realize the adaptive granularity of the content object. Finally, we show a simple concept model for online authoring systems that support the evolution from resource objects to learning objects. ©2010 IEEE.

With the improvement of internet technologies and multimedia resources, traditional learning has been replaced by distance learning, web-based learning or others' e-learning learning styles. According to distance learning, there are many research organizations and companies who make efforts in developing the relevant systems. But they lack interoperability. The only way to reuse these applications is to redevelop them for specific purposes. In order to solve this situation and norm the various learning resources, IMS proposes a new e-learning standard named "Common Cartridge". This standard not only integrates the past e-learning standards like LOM, SCORM and QTI but also proposes a technical architecture called Learning Tools Interoperability to allow applications to reuse different systems without reprogramming. In this paper, we firstly introduce the current e-learning environment. Then we pay attention on the usage of Common Cartridge standards and discuss the architecture of Learning Tools Interoperability. According to these standards, we will point out the e-learning standard that might be widely utilized in the future. Copyright © 2010, IGI Global.

The information resources on the Web are diversified, the amount of which is increasing rapidly. Demands for selecting useful information from the Internet, managing personal contents, and sharing contents under control have risen. In this study, we propose the Ubiquitous Personal Study, a framework of personalized virtual study to support accessing, managing, organizing, sharing and recommending information. In this paper, we focus on discussing the framework, and design and implementation issues on how to implement it with Web 2.0 mash-up technology and Open Source Software.

User profiling is a widely used technique to analyse and store user interests and preferences to apply this knowledge to improve user experiences with information systems. In this research paper, we present an approach for dynamically constructing user profiles, particularly from uniform relevance feedback in information-seeking activities. We propose an inference method for user interests, which we call High-Similarity Sequence Data-Driven (H2S2D) clustering and discuss its peculiarities and show its superiority for the creation of high-quality concepts, which are the elementary constituents of user profiles. To reflect the volatility of user interests and emphasise the steadiness of persistent preferences, we adopt recency, frequency and persistency as the three main criteria for multi-layered dynamic profile construction and update. Copyright © 2009 Inderscience Enterprises Ltd.

User-centeredness has become an essential feature of information systems directly interacting with a user. In this paper we emphasize the importance of a user as an active content creator and contributor, community, and their synergy to achieve better user-centeredness in personalized Web search services. For this, we realize individual-community collaborative scheme integrating information search and sharing, and thus benefiting from both individual and community activities. Here we show the details of the scheme, including expertise-based information co-evaluation and collaborative search.

E-Learning makes it possible for learners to study at any time. With the improvement of ubiquitous technologies and corresponding devices, learners can conduct their learning activities without any limitations from the environment. Instructors can utilize various kinds of resources to create more plentiful learning materials for learners. In this situation, instructors or learners have to spend more time on collecting relevant resources for a specific purpose. In our previous works, we utilized the geographical information and RFID (Radio Frequency Identification) to develop an outdoor adventure game to support learning procedures. In this paper, we go further to propose and develop a resource retrieval mechanism to assist learners in collecting relevant learning resources. This mechanism can be regarded as a search procedure. We aim at providing search results in the order of relevant degree based on learners' current geographical information. It can make resource retrieval more efficient in ubiquitous learning environments.

The paper first introduces the concept of flowable service, the basic idea of which is to organize circumjacent services in such a way so that they are perceived by an individual as those naturally embedded in his/her surrounding environment. It is achieved by seamless integration and provision of diverse services in an intuitive "flowable" way. For this, the mechanism responsible for service discovery, composition and provision has to have a good understanding of a user it serves to. We design a framework with a special focus on context-awareness to find and integrate the circumjacent services thought to be needed by each individual at his/her current situation, and thus create an ambient service environment tailored to his/her particular needs and characteristics. Here introduce the framework, its components and process abstraction, and outline some major issues we have to face when implementing our approach.

In this paper we introduce a novel model called flowable service model to seamlessly integrate diverse services in an intuitive "flowable" way. With this model we seek to achieve maximum satisfaction of both service providers and consumers, and reduce the delivery cost of services at the same time. We describe the principal characteristics of the model and its peculiarities, and illustrate ideas on mediation and integration of flow services inspired by Human Information Processing.

In this study, we propose a distributed framework for flexible and collaborative information sharing. It integrates diverse information media platforms, which are well-known as collaboration spaces where an individual user can share his/her contents with others freely, by using a hybrid peer-to-peer network model. Further, the framework ensures user collaboration on different reconfigurable scales: individual, small-group and large-group. Here we discuss the framework and show its prototype implementation that supports creating, managing, exchanging and sharing information among several different media platforms.

In this study, we propose a Gradual Adaption Model for a Web recommender system. This model is used to track users' focus of interests and its transition by analyzing their information access behaviors, and recommend appropriate information. A set of concept classes are extracted from Wikipedia. The pages accessed by users are classified by the concept classes, and grouped into three terms of short, medium and long periods, and two categories of remarkable and exceptional for each concept class, which are used to describe users' focus of interests, and to establish reuse probability of each concept class in each term for each user by Full Bayesian Estimation as well. According to the reuse probability and period, the information that a user is likely to be interested in is recommended. In this paper, we propose a new approach by which short and medium periods are determined based on dynamic sampling of user information access behaviors. We further present experimental simulation results, and show the validity and effectiveness of the proposed system. © 2009 IEEE.

This research is aimed at a smart music playing environment in which a set of speakers are placed in different locations/rooms and one of the speakers will automatically play a music whenever a user gets close to it. Our study is focused on the design and prototype implementation of such a smart music playing system called U-LAMP (User Location-Aware Music Player). It consists of a server that stores streamed music data, and a set of clients that can receive and play the streamed music from the server. A user carries a RFID tag and can move around. When the user's tag ID is detected by a RFID reader that is connected to a nearby client, the server will switch the transmission of streamed music data from the previous client to the current one. This paper explains in details about how to get a user's location by detecting the user's tag ID, manage RTP-based music data delivery from the server to the client, play a streamed music using JMF, and achieve better audio listening effects during the transition period of handing-over the playing music stream between speakers in different rooms. © 2009 IEEE.

With the rapid advancing in electronics, VLSI, and nanotechnology, ubiquitous devices (u-objects) like mobile devices, sensing devices, RFID devices, IC cards, wireless devices, wireless wearable devices, etc. are getting more and more in number, smaller and smaller in size, and even invisible in form. In order to make them work without human involved and interaction, there is necessity of enabling them with intelligent, smart, or autonomic computing capabilities. Of course, intelligent function embedded is a straightforward approach. However, it may be difficult for the tiny devices to embed the intelligent computing capabilities due to their limited capacity and processing power. For a multiple intelligent agents system, it is not efficient to develop and maintain intelligent computing functions for each individual agent. Therefore, the idea is to design an intelligence entity sharing pool in which a requested intelligence entity can be dynamically composed from a number of atomic intelligence entities and complex intelligence entities residing in the intelligence entity pool and posted to virtual objects (v-object) in the v-object server for sharing by ubiquitous devices in the real world. This research is of three main phases: designing the intelligence entity sharing pool
mapping u-objects to virtual objects (v-objects)
and developing a framework for intelligence entity sharing. This paper is mainly focused on describing the first two phases. © 2009 IEEE.

In this study, we present a gradual adaptation model to estimate human information access behavior. A variety of users' information access data are collected in terms of short, medium, long periods, or by categories such as remarkable and exceptional. The proposed model is then established by analyzing the preprocessed data based on the Full Bayesian Estimation.

With slide-film interface discussed in this paper we propose a new mobile Web search experience. The interface abolishes fatigue-inducing scrolling while preserving "quality" summaries of Web search results to improve the efficacy and efficiency of mobile Web search. The evaluation of the approach shows that the proposed interface is superior to the conventional mobile search method in a number of aspects like easy navigation and good viewability of search results and can be its good alternative.

Relevance feedback is an important source of information about a user and often used for usage and user modeling for further personalization of user-system interactions. In this paper we present a method to infer the user's interests from his/her relevance feedback using an online incremental clustering method For inference of a new interest (concept) and concept update the method uses the similarity characteristics of uniform user relevance feedback It is fast, easy to implement and gives reasonable clustering results. We evaluate the method against two different data sets, demonstrate and discuss the outcomes.

In the situation of information overload we are experiencing today, conventional web search systems taking a one-size-fits-all approach are often not capable of effectively satisfy individual information needs. To improve the quality of web information retrieval, we propose a collaborative personalised search approach that makes an attempt to &apos;understand&apos; and better satisfy the information needs for each and every searching user. We present a web information retrieval framework called Better Search and Sharing (BESS) that Captures user-system interactions, profiles them and induces personal interests that changes over time with an interest-change-driven profiling mechanism that is also extensively used for the co-evaluation of documents found valuable inside a specific search context by users with similar interests.

The information resources on the Web are diversified, and the amount is increasing rapidly. Demands for selecting useful information from the Internet, managing personal contents, and sharing contents under control have risen. In this study, we propose the Ubiquitous Personal Study (UPS), a framework of personalized virtual study to support accessing, managing, organizing, sharing and recommending information. In this paper, we focus on discussing the issues on how to implement it with Web 2.0 mash-up technology and Open Source Software.

It is well known that alongside with search engine performance improvements and functionality enhancements one of the determinant factors of user acceptance of any search service is the interface. This factor is particularly important for mobile Web search mostly due to small screen limitations of handheld devices. In this paper we propose scrolless mobile Web search interface to decrease search efforts that are multiplied due to these limitations, and discuss its potential advantages and drawbacks over conventional one.

A smart space is a physical spatial environment such as a room that can provide automatic responses according to the contextual information occurring in an environment. The context data is usually acquired via various devices which are installed somewhere in the environment and/or embedded into some special physical objects, called u-objects, which are capable of executing computation and/or communication. However, the devices used in current researches on smart space are either fixed or residing in real. objects, which are incapable of moving by themselves. Likewise, more and more robots have been produced, but most of them are designed and developed based on the assumption that the space surrounding a robot is an ordinary one, i.e., a non-smart space. Therefore, it is necessary to study what additional services can be offered and what specific technical issues will be faced when adding robots to smart spaces. To identify the potential novel services and technology problems, this research is focused on a case study of a u-object finding service performed by a robot in a smart space. This paper presents the design and development of the system prototype robot which can communicate with other devices and can find a u-object with attached RFID tag in a smart room. Some preliminary experiments were conducted and the results verified the functionalities of the system.

Lifelog is a data set composed of one or more media forms that record the same individual's daily activities. One of the main challenging issues is how to extract meaningful information from the huge and complex lifelog data which is continuously captured and accumulated from multiple sensors. This study is focused on the activity models and analysis techniques to process lifelog data in order: to find what events/states are interesting or important, to summarize the useful records in some structured and semantic ways for efficient retrievals and presentations of past life experiences, and to use these experiences to further improve the individual's quality of life. We propose an integrated technique to process the lifelog data using the correlations between different kinds of captured data from multiple sensors, instead of dealing with them separately. To use and test the proposed models and the analysis techniques, several prototype systems have been implemented and applied to some domain-specific lifelog data; such as in improving a group's collaborative efforts in revising a software, in managing kid's outdoor safety care, in providing a runner's workout assistance, and in structuring lifelog image generation, respectively.

Knowing each user's information needs is important for information systems to better facilitate human information activities. This is especially important in the days of information overload we are experiencing today. However, knowing and correctly applying individual information needs is extremely difficult, often impossible. Yet knowing multiple contexts of user information behavior can give us some conception (or a hint) of conceivable information a user tries to obtain in a particular context. In this paper we focus on capturing such information contexts into dynamically changing profiles which are further used to facilitate the user's information seeking activities. © 2008 IEEE.

In this study, we propose a gradual adaption model for estimation of user information access behavior. A variety of users' information access data are collected by unit of a day for each user, and analyzed in terms of short, medium, long periods, and by remarkable and exceptional categories. The proposed model is then established by analyzing the pre-processed data based on Full Bayesian Estimation. We further present experimental simulation results, and show the operability and effectiveness of our proposed model. © 2008 IEEE.

We discuss an approach to enhance conventional IR with contributions (relevance judgments) done by users interacting with a document collection and personalization by profiles inside the collaborative search and sharing system. It is designed to capture user search intentions derived from explicit and implicit feedback and keep analyzed data in layered profiles for further query augmentation, document recommendation and contribution personalization in the form of "subjective index." For profile generation and document evaluation we propose an interest change driven profiling mechanism which reflects all dynamics of user search intentions. © 2007 IEEE.

In recent years, a variety of new computing paradigms have been proposed for various purposes. It is true that many of them intend to and really can gratify some of the people sometime, somewhere; a few of them can even gratify some of the people anytime, anywhere. However, at present, none of the computing paradigms intend to gratify all the people anytime, anywhere. With the rapid advance of information technology and the spread of information services, the IT disparity in age, social standing, and race of the people has been expanding and has become a critical social problem of the 21st century. Thus, we have a fundamental question: Can we construct, in a unified methodology, a computing environment that can gratify all the people in all situations, all places and all the time? We propose a novel and inclusive computing paradigm, named ubisafe computing, for studying and providing possible solutions to the above problem. The ultimate goal of ubisafe computing is to build a computing environment in which all people and organizations can benefit from ubiquitous services anytime anywhere with assured and desired satisfaction without worrying or thinking about safety. That is, the ubisafe computing vision emphasizes two basic aspects: ubiquitous safety and ubiquitous satisfaction to all people in all situations. This paper presents the motivations for the ubisafe computing vision but focuses on one basic aspect of ubiquitous safety that covers reliability, security, privacy, persistency, trust, risk, out of control, and other watchfulness while considering novel, essential ubicomp or percomp features of unobtrusive computers, diverse users/people and life-like systems.

We proposed a collaborative information sharing environment based on P2P net-working technology, to support communication among special groups with given tasks, ensure fast information exchange, increase the productivity Of working groups, and reduce maintenance and administration costs in our previous work.
However, for a social growing community, not only the information exchange/sharing functions are necessary, but also solutions to support users with idea and knowledge publication tools for private purpose or public use are essential. Some private message (personal idea and experience) posting tools (e.g., weblog) and group collaborative knowledge editing tools (e.g., Wikis) are used in practice; the merits of these tools have been recognized.
In this paper we propose a scalable information sharing solution, which integrates decentralized P2P networking with centralized personal/group media tools. This solution combines the effective tools, such as weblog and Wiki, into P2P-based collaborative groupware system, to facilitate infinite, growing and scalable information management and sharing for individuals and groups.

Context is the information, which is created and obtained from the surrounding environment for the interaction between humans and computational services. A generic model is a key accessor to the context in any context-aware applications for ubiquitous computing. In the past decades, a number of context modeling techniques have been proposed e.g. markup scheme based, logic-based, graphical, and ontology-based. Since ontology in its nature is a promising tool to specify concepts and interrelations, it has been widely adopted in context modeling. However, in the rapid changing environments, semantics may vary according to the time factors and dynamic group of users. In this paper, we propose an ontology-based context model with temporal vector space in order to complement this deficiency.

Nowadays, the World Wide Web offers public search services by a number of Internet search engine companies e.g. Google [16], Yahoo! [17], etc. They own their internal ranking algorithms, which may be designed for either general-purpose information and/or specific domains. In order to fight for bigger market share, they have developed advanced tools to facilitate the algorithms through the use of Relevance Feedback (RF) e.g. Google's Toolbar. Experienced by the black-box tests of the RF toolbar, all in all, they can acquire simple and individual RF contribution. As to this point, in this paper, we have proposed a collaboratively shared Information Retrieval (IR) model to complement the conventional IR approach (i.e. objective) with the collaborative user contribution (i.e. subjective). Not only with RF and group relevance judgments, our proposed architecture and mechanisms provide a unified way to handle general purpose textual information (herein, we consider e-Learning related documents) and provide advanced access control features [15] to the overall system.

Distance education, e-learning, and virtual university are similar terms for a trend of modern education. It is an integration of information technologies, computer hardware systems, and communication tools to support educational professionals in remote teaching. This chapter presents an overview of distance education from the perspective of policy, people, and technology. A number of questions frequently asked in distance learning panel discussions are presented, with the suggested answers from the authors. The survey presented in this chapter includes communication, intelligent, and educational technologies of distance education. Readers of this chapter are academic researchers and engineers who are interested in new research issues of distance education, as well as educators and general participants who are seeking for new solutions. © 2007, Idea Group Inc.

Search is not the only activity in information space users are engaged in. Most of time users rediscover things they used to find in the past, and often they just browse without any specific purpose discovering information space around them or with a particular purpose, such as learning miscellaneous information. We propose an approach for designing exploratory information space that makes use of human-centered power of bookmarking for information selection. The information space is built as a result of a search for something a user intends to discover, and serves as a place for rediscoveries of personal findings, socialization and exploration inside discovery chains of other participants of the proposed prototype system. The most important elements of the proposed information space, main elements of the prototype system to be implemented and used to explore the space, and use scenarios are discussed here. © 2006 IEEE.

In this study, we propose and develop an opensource groupware system called NetIsle. NetIsle is a general purpose groupware system for uniform open groups that integrate a number of tools for online collaboration to ensure fast information exchange and sharing, increase the productivity of working groups, and reduce maintenance and administration costs. The main technologies used for the construction of the system are peer-to-peer (P2P) and push, which are best fitted to those principles and beliefs we build our system upon. © Emerald Group Publishing Limited.

This paper outlines anew model of ubiquitous computing technology support learning (uLearning). In the uLearning, in order to facilitate the awareness of other helpful learners, increase the availability of the mutual and friendly support in the learning process, and enhance learners' communication skills, the authors offer a serial of services. These services arc designed as dynamic group construction service, social intercommunion facilitation service, service of navigating learner to get out of difficulty, etc.

With advanced technologies, computer devices have become smaller and powerful. As a result, many people enjoy ubiquitous learning using mobile devices such as Pocket PCs. Pocket PCs are easy to carry and use as a distance learning platform. In this paper, we focus on the issues of transferring the current PC based Sharable Content Object Reference Model (SCORM) to Pocket PC based. We will also introduce the Pocket SCORM Run-Time Environment (RTE) which has been developed in our lab. Pocket SCORM architecture is able to operate, even when the mobile device is off-line. It will keep the students' learning record. When it is on line, the records will then be sent back to Learning Management System (LMS). With memory limitation, we provides course caching algorithm to exchange the course content on the Pocket SCORM.

This paper discusses a fundamental notion and model of ubiquitous computing technology support learning (uLearning) system. In order to increase learners' social skills and support social interaction among learners, the authors design a collection of support tools; these tools serve for three processes: encounter communication and collaboration. Furthermore, prototyping implementation of the social interaction support tools is introduced.

This paper describes the computer supported ubiquitous learning system and the social interaction between learners. We define the ubiquitous learning with five prime attributes, and present a generalized social interaction support model for the ubiquitous learning. Moreover in order to support learners with increasing social skill, a solution for constructing social interaction in ubiquitous learning environment is designed, which includes three major functions: encounter communication and collaboration support functions.

In this study, we propose and develop an open-source groupware system called NetIsle. NetIsle is a general-purpose groupware system for uniform open groups that integrate a number of tools for online collaboration to ensure fast information exchange and sharing, increase the productivity of working groups, and reduce maintenance and administration costs. The main technologies used for the construction of the system are peer-to-peer (P2P) and push, which are best fitted to those principles and beliefs we build our system upon.

This paper presents a multi-agent system to assist a teacher managing his/her course contents placed on web servers. In this system there is a set of agents and every agent may work independently from or collaboratively with others. Once generated, an agent can reside in a teacher's daily working computer (called administration host) or a proxy host, and can move between the two hosts. Each agent is devoted to one piece of job and all of them, as a whole, coordinately conduct a sequence of management work during the entire process of teaching a course. A teacher may administrate agents via a specific system shell on his/her administration host, or a usual web browser on another computer/PDA/mobile phone. The system has been carefully modularized, and thus a new type of agent, if necessary, can be relatively easily developed and quickly incorporated into the system to further enhance or extend its management capability.

This paper describes small group collaboration solutions based upon push technology, which provides awareness for collaboration, reduces efforts for necessary information discovery, and makes users collaboration initiators, active information providers and new values creators. The push technology-based tools, i.e. File Pusher, Scheduler, NetIsle Mailer, etc., axe implemented in a hybrid peer-to-peer general-purpose groupware system, called NetIsle, which has been developed by using Java RMI and sockets and can be run on any platform where Java VM is available.

In recent years, we have encountered a serious problem that a huge amount of information is flooding on the Internet. There is a growing desire to have an effective way to utilize information on the Internet. We propose a framework for collecting, organizing and reusing information through group collaboration: collaborative information browser. We further discuss the design and implemention of a prototype for a collaborative information browser.

Due to the popularization of Internet and World Wide Web (WWW), the limitation of distance and region are broken for business behaviors. At same time, E-Commerce becomes one of the most important applications on Internet E-Commerce can help a company or enterprise to extend its trading area to unlimited space. Some techniques and facilities are proposed to enhance E-Commerce applications. Agent technique is one of the important technologies developed to support the Internet applications. Especially, the Internet and WWW technologies broken the limitation of space of enterprise marketing, and the agent techniques solve the problems of temporality. Because of when the users are off-line, the agents are still active in the world of computer network and play the role of their users. In this paper, agents and mobile agents mechanism are proposed for electronic marketplaces. There are some researching issues will be discussed. They include the platform of mobile agents, the types and classifications of agents and mobile agents, authentication and security facilities, and so on. Based on this architecture of E-marketplace, the applications of E-commerce will be more effective, easier to develop, and more creating the marketing of business.

In this study, we are trying to develop a general-purpose groupware system for uniform open groups that integrate in itself all main tools necessary for collaboration and cooperation online to achieve common goals at shorter times, ensure fast information transfer and decrease the amount of necessary administration. The system consists of a number of tools that reflect actions of one member at all the nodes of the system, giving awareness for collaboration and increasing its effectiveness. In comparison with other applications that have been designed for knowledge sharing, our system pushes necessary information through the network so that no manual search and download are necessary if one of the group members possesses it. Thus the speed of training, knowledge and information sharing is accelerated and participation of the group members will be increased. By developing the system, we are trying to provide shared workspace for all main platforms as well.

In this paper we introduce a framework for every-netizen online learning communities that widely open to anyone who is willing to learn and share knowledge with others across the networks. We discuss how to provide individualized collaboration support for e-learning in the community, and demonstrate a prototype of implementation in a web-based social virtual environment with two supporting tools.

As a new distributed computing paradigm, peer-to-peer (P2P) networks and systems have attracted more and more attentions in recent years. This study aims at proposing and constructing a new information resource sharing system based on the P2P paradigm. In this paper, we focus on solving and improving the hit rate and search speed problems that exist in Gnutella, one of the most popular P2P file sharing systems. We discuss how to introduce XML to describe the attributes and properties of each shared file and integrate an embedded database engine to increase the search speed. Preliminary experiment results have demonstrated that the approach proposed in this study is efficient and useful.

In this paper, we propose a conceptual framework for every-citizen learning communities based on a recently widespread Internet tool known as Multi-user dimension Object-Oriented (MOO). We discuss the design and development of a prototype system of the virtual community based interactive learning environment, which supports human-human communication in addition to human-computer communication. with emphasis on social interaction. (C) 2002 Elsevier Science Inc. All rights reserved.

Community-based collaborative virtual universities that widely open to anyone who has the will to learn and to share their knowledge with others across the networks have been proposed, which imply the diversity of their users, encompassing the needs of people of all ages, nationalities and ability levels, and consequently cause a language problem in knowledge representation and communication. This paper proposes a multilingual agent and discusses how to design the agent in the context of multi agent paradigm to provide an integrated solution of software agent and virtual environment media technologies, with a demonstration of a prototyping implementation in a MOO (Multi user dimension Object-Oriented) based virtual environment.

This paper describes a shopping negotiation agent that can adapt user preferences and automatically negotiate with its counter party on behalf of a user it represents. The agent is built on a basis of the proposed negotiation model, the enhanced extended Bazaar model, which is a sequence of decision making model of negotiation with exploiting common knowledge, public information, and game theory. Since different users can have different preferences, it is important for the agent to have adaptation to different user preferences. This can be achieved by acquiring user preferences, tracing user’s behavior on Web and mapping the behavior to a set of the preference parameters, creating the negotiation model class, and generating an instance negotiation model object with new/updated preference parameters.

The exploration of social networks is essential for finding capable cooperators who can help problem-solving and for augmenting cooperation between workers in an organization. This paper describes PeCo-Mediator-II to seek capable cooperators through a chain of personal connections (PeCo) in a networked organization. Moreover, this system helps to gather, explore, and visualize social networks in an organization. The experimental results show that the system facilitates users' encounters with cooperators and develops new helpful connections with the cooperators.

We determine the optimum time T_OPT to order a spare part for a system before the part in operation has failed. T_OPT is a function of the part's failure-time distribution, the lead(delivery)time of the part, its inventory cost, and the cost of downtime while waiting delivery. The probabilities of the system's up and down states are obtained from a non-regenerative stochastic Petri net. T_OPT is found by minimizing E[cost], the expected cost of inventory and downtime. Three cases are compared:1)Exponential order and lead times, 2)Deterministic order time and exponential lead time, and 3)Deterministic order and lead times. In Case 1, it is shown analytically that, depending on the ratio of inventory to downtime costs, the optimum policy is one of three:order a spare part immediately at t=0, wait until the part in operation fails, or order before failure at T_OPT>0. Numerical examples illustrate the three cases.

In this paper, we propose a conceptual framework for every-citizen learning comminities based on a recently widespread Internet tool known as MOO (Multi user dimension Object-Oriented). We discuss the design and development of a prototype system of virtual community based interactive learning environments, which supports human-human communication in addition to human-computer communication, with emphsis on social interaction.

The exploration of social networks is essential to find capable cooperators who can help in problem-solving and in augmenting cooperation among workers in an organization. PeCo-Mediator-II is an agent based system that helps gathering, exploring, and visualizing social networks among users. This paper proposes a user model with history logs to support exploration, to share individual knowledge, and to reduce overload of cooperators. Experimental results show that the system facilitated users encounter cooperators and helped develop new relationships between users and cooperators.

In the past few years, a network-accessible, multi-user simulated environment known as MUD has becoming widespread on the Internet for purposes of entertainment and education. In this paper, we firstly discuss design issues of collaborative learning in text-based virtual reality environments. We describe our experimental MUD based virtual learning environment, its integration with outside databases and Internet resources, and a graphical user interface to use it. Finally, we introduce our experimental use of this virtual learning environment, experiences from having virtual lessons in it, and some primary evaluation results.

Petri nets can be viewed as a kind of language for specifying Markov models of a system, because of their highly visual nature that can give insight into the nature of the modeled system. In this paper, an aggregate approach by extended stochastic Petri net and Markov renewal process with some non-regeneration points is proposed to conduct performance analysis of an Ethernet-based flexible manufacturing system. The proposed approach may improve efficiency of performance analysis for most practical distributed network based manufacturing systems.

In the past few years, a network-accessible, multi-user simulated environment known as MUD has becoming widespread on the Internet for purposes of entertainment and education. In this paper, we firstly discuss design issues of collaborative learning in text-based virtual reality environments. We describe our experimental MUD based virtual learning environment, its integration with outside databases and Internet resources, and a graphical user interface to use it. Finally, we introduce our experimental use of this virtual learning environment, experiences from having virtual lessons in it, and some primary evaluation results.

We develop a new class of stochastic Petri net. non-regenerative stochastic Petri net (NRSPN), which allows the firing time of its transitions with arbitrary distributions, and can automatically generate a bounded reachability graph that is equivalent to a generalization of the Markov renewal process in which some of the states may not constitute regeneration points. Thus, it can model and analyze behavior of a system whose states include some non-regeneration points. We show how to model a system by the NRSPN, and how to obtain numerical solutions For the NRSPN model. The probabilistic behavior of the modeled system can be clarified with the reliability measures such as the steady-state probability, the expected numbers of visits to each slate per unit lime, availability, unavailability and mean time between system failure. Finally, to demonstrate the modeling ability and analysis power of the NRSPN model, we present an example for a fault-tolerant system using the NRSPN and give numerical results for specific distributions.

When interacting and acting in a dynamic multiagent environment, an agent must ask itself what is presently its best course of action given what it now knows about what the environment will be like when it intends to act. It requires some ability to make decision by computing the probability that relevant propositions will hold at a specified point of time. Among the interaction and action, some of them can be characterized by an exponential time distribution, others might not. This paper presents a non-regenerative stochastic Petri net that allows arbitrary time distributions, use it to model and analyze agents' probabilistic behavior.