Recommending Personalized POIs from Location Based Social Network

Location based social networks（ LBSNs） provide location specific data generated from smart phone into online social networks thus people can share their points of interest（ POIs）. POI collections are complex and can be influenced by various factors,such as user preferences,social relationships and geographical influence. Therefore,recommending new locations in LBSNs requires to take all these factors into consideration. However,one problem is how to determine optimal weights of influencing factors in an algorithm in which these factors are combined. The user similarity can be obtained from the user check-in data,or from the user friend information,or based on the different geographical influences on each user＇s check-in activities. In this paper,we propose an algorithm that calculates the user similarity based on check-in records and social relationships,using a proposed weighting function to adjust the weights of these two kinds of similarities based on the geographical distance between users. In addition,a non-parametric density estimation method is applied to predict the unique geographical influence on each user by getting the density probability plot of the distance between every pair of user＇s check-in locations. Experimental results,using foursquare datasets,have shown that comparisons between the proposed algorithm and the other five baseline recommendation algorithms in LBSNs demonstrate that our proposed algorithm is superior in accuracy and recall,furthermore solving the sparsity problem.

PECS: Towards Personalized Edge Caching for Future Service-Centric Networks

Mobile operators face the challenge of how to best design a service-centric network that can effectively process the rapidly increasing number of bandwidth-intensive user requests while providing a higher quality of experience(QoE). Existing content distribution networks(CDN) and mobile content distribution networks(mCDN) have both latency and throughput limitations due to being multiple network hops away from end-users. Here, we first propose a new Personalized Edge Caching System(PECS) architecture that employs big data analytics and mobile edge caching to provide personalized service access at the edge of the mobile network. Based on the proposed system architecture, the edge caching strategy based on user behavior and trajectory is analyzed. Employing our proposed PECS strategies, we use data mining algorithms to analyze the personalized trajectory and service usage patterns. Our findings provide guidance on how key technologies of PECS can be employed for current and future networks. Finally, we highlight the challenges associated with realizing such a system in 5G and beyond.

A Personalized Search Model Using Online Social Network Data Based on a Holonic Multiagent System

Personalized search utilizes user preferences to optimize search results,and most existing studies obtain user preferences by analyzing user behaviors in search engines that provide click-through data.However,the behavioral data are noisy because users often clicked some irrelevant documents to find their required information,and the new user cold start issue represents a serious problem,greatly reducing the performance of personalized search.This paper attempts to utilize online social network data to obtain user preferences that can be used to personalize search results,mine the knowledge of user interests,user influence and user relationships from online social networks,and use this knowledge to optimize the results returned by search engines.The proposed model is based on a holonic multiagent system that improves the adaptability and scalability of the model.The experimental results show that utilizing online social network data to implement personalized search is feasible and that online social network data are significant for personalized search.

Perspectives for Sharing Personal Information on Online Social Networks

The goal of this research was to study how people feel about sharing personal information on social networks. The research was done by interviews;50 people were interviewed, mostly from China's Mainland, Hong Kong, and Finland. This paper presents the included 12 questions and discusses the collected answers. It was discovered, e.g., that 38 out of the 50 answerers use social media every day and share versatile personal information on the Internet. Half of the answerers also share information about other people on the Internet. It was also discovered that compared to male answerers, the female answerers were more active in sharing information about other people. There was a significant variety in opinions: what should be the age limit for sharing personal information online, while 22 out of the 50 answerers felt that there is no need for an age limit at all. According to the answers, only a few people use social media for making new friends. Instead, an important reason for using social media is that their existing friends are using. An interesting finding was that the answerers see the Internet as a part of the real world;the privacy that you have on the Internet is the privacy that you have in the real world.

All-To-All Personalized Communication in Wormhole-Routed 2D/3D Meshes and Multidimensional Interconnection Networks

All-to-all personalized communication,or complete exchange,is at the heart of numerous applications in paral-lel computing.It is one of the most dense communication patterns.In this paper,we consider this problem in a2D/3D mesh and a multidimensional interconnection network with the wormhole-routing capability.We propose complete ex-change algorithms for them respectively.We propose O(mn 2 )phase algorithm for2D mesh P m ×P n and O(mn 2 l 2 )phase algo-rithm for3D mesh P m ×P n ×P l ,where m,n,l are any positive integers.Also O(ph(G 1 )n 2 )phase algorithm is proposed for a multidimensional interconnection network G 1 ×G 2 ,where ph(G 1 )stands for complete exchange phases of G 1 and|G 2 |=n.

基于联邦学习的聚酯纤维酯化过程温度预测研究

在聚酯纤维生产过程中,酯化温度的精确控制至关重要。然而,由于生产设备和工艺参数的差异,传统预测方法难以满足个性化需求,且数据共享过程中存在隐私泄露和通信压力问题。提出了一种基于联邦学习的个性化自适应酯化温度时间序列预测算法。采用联合预测机制为每个客户端分配私有和共享模型,设置自适应阶段根据客户端数据分布智能调整模型参数,并学习客户端独有的联合预测权重,实现个性化预测输出。采用贝叶斯优化算法解决高维复杂和资源限制问题,快速高效地得到了最佳超参数组合。在三家聚酯纤维生产厂家的真实数据集上的广泛实验结果表明,算法在五种预测模型下均取得了最佳预测性能,有效提高了酯化温度预测的准确性。

Ney man-Person准则的神经网络实现新算法(英文)

假设检验中Neyman_Person准则是一种基于似然比的信号分类、检测、识别方法 .神经网络是实现这种判定准则的优选方案 ,但是传统的最小平方学习算法 ,如BP算法等 ,往往不能取得全局最优解 .本文针对一种非最小平方学习算法 ,提出了一种概率分配原则 ,并给出了一种Neyman_Person准则的神经网络实现新算法 .文中对新算法在假设检验中的应用进行了仿真验证 ,结果表明新算法具有更小的误差 ,更加适用于Neyman_Person准则 .

基于个性化PageRank和对比学习的图异常检测模型

图异常检测旨在从属性网络中检测出异常节点,其由于在许多应用领域如金融、电子贸易、垃圾邮件发送者检测中有着深远的实际意义而备受重视。传统的非深度学习方法只能捕捉图的浅层结构,对此,研究者们提出了基于深度神经网络的异常检测模型。然而,这些模型没有考虑到图中节点的中心性差异,这种差异在捕获节点的局部信息时会导致信息缺失或引入远端节点的噪声。此外,它们忽略了属性空间的特征信息,这些信息可以提供额外的异常监督信号。为此,从无监督的视角出发,提出了一种新颖的基于个性化PageRank和对比学习的图异常检测框架PC-GAD(Personalized PageRank and Contrastive Learning based Graph Anomaly Detection)。首先,提出一种动态采样策略,即通过计算图中每个节点的个性化PageRank向量确定其相应的子图采样数目,避免局部信息的缺失和引噪;其次,针对每个节点,分别从拓扑结构和属性空间的角度出发捕获节点的异常监督信号,并设计相应的对比学习目标,从而全面地学习潜在的异常模式;最后,经过多轮对比预测,根据输出的异常值得分评估每个节点的异常程度。为验证所提模型的有效性,分别在6个真实数据集上与基准模型开展了大量对比实验。实验结果验证了PC-GAD能够全面地识别出图中的异常节点,AUC值相比现有模型提升了1.42%。

Egocentric social network correlates of physical activity

Background:The social environment might play an important role in explaining people’s physical activity(PA)behavior.However,little is known regarding whether personal networks differ between physically active and physically inactive people.This study aimed to examine the relationship between personal network characteristics and adults’physical(in)activity.Methods:An egocentric social network study was conducted in a random sample in Switzerland(n=529,mean age of 53 years,54%females).Individual and personal network measures were compared between regular exercisers and non-exercisers.The extent of these factors’association with PA levels was also examined.Results:Non-exercisers(n=183)had 70%non-exercising individuals in their personal networks,indicating homogeneity,whereas regular exercisers(n=346)had 57%regularly exercising individuals in their networks,meaning more heterogeneous personal networks.Additionally,having more regular exercisers in personal networks was associated with higher PA levels,over and above individual factors.Respondents with an entirely active personal network reported,on average,1 day of PA more per week than respondents who had a completely inactive personal network.Other personal network characteristics,such as network size or gender composition,were not associated with PA.Conclusion:Non-exercisers seem to be clustered in inactive networks that provide fewer opportunities and resources,as well as less social support,for PA.To effectively promote PA,both individuals and personal networks need to be addressed,particularly the networks of inactive people(e.g.,by promoting group activities).

网络融合背景下的电视IP化趋势

在网络融合的背景下,电信网、宽带通信网、数字电视网与互联网的融合日益加深,极大推动了电视产业的变革。电视IP化作为该变革的重要方向,能够有效推动电视产业的创新发展。基于此,本文对电视IP化的内涵与特征以及网络融合背景下电视IP化的重要性进行简要分析,探讨了网络融合推动电视IP化的发展路径,以期为电视产业的转型升级提供参考。

Critical Factors for Personal Cloud Storage Adoption in China

Purpose： In order to explain and predict the adoption of personal cloud storage, this study explores the critical factors involved in the adoption of personal cloud storage and empirically validates their relationships to a user＇s intentions.Design/methodology/approach： Based on technology acceptance model（TAM）, network externality, trust, and an interview survey, this study proposes a personal cloud storage adoption model. We conducted an empirical analysis by structural equation modeling based on survey data obtained with a questionnaire.Findings： Among the adoption factors we identified, network externality has the salient influence on a user＇s adoption intention, followed by perceived usefulness, individual innovation, perceived trust, perceived ease of use, and subjective norms. Cloud storage characteristics are the most important indirect factors, followed by awareness to personal cloud storage and perceived risk. However, although perceived risk is regarded as an important factor by other cloud computing researchers, we found that it has no significant influence. Also, subjective norms have no significant influence on perceived usefulness. This indicates that users are rational when they choose whether to adopt personal cloud storage.Research limitations： This study ignores time and cost factors that might affect a user＇s intention to adopt personal cloud storage.Practical implications： Our findings might be helpful in designing and developing personal cloud storage products, and helpful to regulators crafting policies.Originality/value： This study is one of the first research efforts that discuss Chinese users＇ personal cloud storage adoption, which should help to further the understanding of personal cloud adoption behavior among Chinese users.

An Application-Oriented Network Model for Wireless Sensor Networks

Wireless sensor networks (WSNs) are energy-constrained networks. The residual energy real-time monitoring (RERM) is very important for WSNs. Moreover, network model is an important foundation of RERM research at personal area network (PAN) level. Because RERM is inherently application-oriented, the network model adopted should also be application-oriented. However, many factors of WSNs applications such as link selected probability and ACK mechanism etc. were neglected by current network models. These factors can introduce obvious influence on throughput of WSNs. Then the energy consumption of nodes will be influenced greatly. So these models cannot characterize many real properties of WSNs, and the result of RERM is not consistent with the real-world situation. In this study, these factors neglected by other researchers are taken into account. Furthermore, an application-oriented general network model (AGNM) for RERM is proposed. Based on the AGNM, the dynamic characteristics of WSNs are simulated. The experimental results show that AGNM can approximately characterize the real situation of WSNs. Therefore, the AGNM provides a good foundation for RERM research.

Identifying Biomarkers for Diabetic Kidney Disease Using GraphSAGE Neural Network

Diabetic Kidney Disease (DKD) is a common chronic complication of diabetes. Despite advancements in accurately identifying biomarkers for detecting and diagnosing this harmful disease, there remains an urgent need for new biomarkers to enable early detection of DKD. In this study, we modeled publicly available transcriptome datasets as a graph problem and used GraphSAGE Neural Networks (GNNs) to identify potential biomarkers. The GraphSAGE model effectively learned representations that captured the intricate interactions, dependencies among genes, and disease-specific gene expression patterns necessary to classify samples as DKD and Control. We finally extracted the features of importance;the identified set of genes exhibited an impressive ability to distinguish between healthy and unhealthy samples, even though these genes differ from previous research findings. The unexpected biomarker variations in this study suggest more exploration and validation studies for discovering biomarkers in DKD. In conclusion, our study showcases the effectiveness of modeling transcriptome data as a graph problem, demonstrates the use of GraphSAGE models for biomarker discovery in DKD, and advocates for integrating advanced machine-learning techniques in DKD biomarker research, emphasizing the need for a holistic approach to unravel the intricacies of biological systems.

Unweighted Voting Method to Detect Sinkhole Attack in RPL-Based Internet of Things Networks

The Internet of Things(IoT)consists of interconnected smart devices communicating and collecting data.The Routing Protocol for Low-Power and Lossy Networks(RPL)is the standard protocol for Internet Protocol Version 6(IPv6)in the IoT.However,RPL is vulnerable to various attacks,including the sinkhole attack,which disrupts the network by manipulating routing information.This paper proposes the Unweighted Voting Method(UVM)for sinkhole node identification,utilizing three key behavioral indicators:DODAG Information Object(DIO)Transaction Frequency,Rank Harmony,and Power Consumption.These indicators have been carefully selected based on their contribution to sinkhole attack detection and other relevant features used in previous research.The UVM method employs an unweighted voting mechanism,where each voter or rule holds equal weight in detecting the presence of a sinkhole attack based on the proposed indicators.The effectiveness of the UVM method is evaluated using the COOJA simulator and compared with existing approaches.Notably,the proposed approach fulfills power consumption requirements for constrained nodes without increasing consumption due to the deployment design.In terms of detection accuracy,simulation results demonstrate a high detection rate ranging from 90%to 100%,with a low false-positive rate of 0%to 0.2%.Consequently,the proposed approach surpasses Ensemble Learning Intrusion Detection Systems by leveraging three indicators and three supporting rules.

结合自我特征和对比学习的推荐模型

针对图神经网络推荐中图卷积在消息传递过程的嵌入表示过平滑和噪声问题,提出一种结合自我特征和对比学习的推荐模型(SfCLRec)。采用预训练-正式训练架构训练模型,首先预训练用户和项目的嵌入表示,通过融合节点自我特征维持节点本身的特征唯一性,并引入层级对比学习任务减少来自高阶邻居节点中的噪声;其次,在正式训练阶段根据评分机制重新构建协同图邻接矩阵;最后,根据最终嵌入得到预测评分。实验结果表明,相较于LightGCN、SimGCL(Simple Graph Contrastive Learning)等现有图神经网络推荐模型,SfCLRec在3个公开数据集ML-latest-small、Last.FM和Yelp中均取得了较好的召回率和归一化折损累计增益(NDCG),验证了SfCLRec的有效性。

一种基于路网位置混淆的个性化位置隐私保护方案

为了解决车载用户在路网中使用基于位置服务造成的位置隐私泄露问题,提出一种基于混淆的个性化位置隐私保护OPLPP方案.首先,基于差分隐私定义一个新的路网不可区分性RN-I概念,以此来衡量路网中位置的不可区分程度;其次,考虑到用户在不同位置的隐私要求,提出个性化隐私预算PPB算法,使用距离占比作为衡量指标,为不同位置分配个性化的隐私预算;最后,利用RN-I设计了连接点-区间混淆C-IO算法,该算法由连接点扰动和区间扰动组成,以此来混淆车辆在道路上的位置.实验结果表明,所提出的OPLPP方案优于现有的二维位置混淆方案,能够在保护用户位置隐私的前提下,提供更高的服务质量.

基于PathSim的MOOCs知识概念推荐模型

大规模开放在线课程提供大规模开放式在线学习平台,为推进现代教育发挥关键作用。然而,减少用户学习盲区和改善用户体验方面的研究仍具有挑战性:交互数据稀疏;难以扩展到大型推荐任务上;用户需求不单由用户喜好决定,还受到不同教师、课程影响;以统一的方式对课程学习事件中不同类型实体及关系进行建模并不妥靠。基于此,引入相关性度量,依据全图结构信息计算各边权重,提出采用相关性度量算法PathSim进行邻域采样的知识概念推荐模型PathSimSage。各实体间相关性得分可在本地离线计算,将神经网络与传播过程分离,保证神经网络的堆叠层数和传播过程的独立性,大幅减少模型所需训练时间。在公开的MoocCube数据集上进行了综合实验,PathSimSage降低了不相关的信息甚至噪声的影响,解决随机游走采样所引发的高度节点偏差问题,并在一定程度上缓解了过平滑效应。

Privacy disclosure in Chinese social networking sites:An investigation of Renren users

Purpose:This study was conducted to investigate the current situation of privacy disclosure(in the Chinese social networking sites.Design/methodology/approach:Data analysis was based on profiles of 240 college students on Renren.com,a popular college-oriented social networking site in China.Users’ privacy disclosure behaviors were studied and gender difference was analyzed particularly.Correlation analysis was conducted to examine the relationships among evaluation indicators involving user name,image,page visibility,message board visibility,completeness of education information and provision of personal information.Findings:A large amount of personal information was disclosed via social networking sites in China.Greater percentage of male users than female users disclosed their personal information.Furthermore,significantly positive relationships were found among page visibility,message board visibility,completeness of education information and provision of personal information.Research limitations:Subjects were collected from only one social networking website.Meanwhile,our survey involves subjective judgments of user name reliability,category of profile images and completeness of information.Practical implications:This study will be of benefit for college administrators,teachers and librarians to design courses for college students on how to use social networking sites safely.Originality /value:This empirical study is one of the first studies to reveal the current situation of privacy disclosure in the Chinese social networking sites and will help the research community gain a deeper understanding of privacy disclosure in the Chinese social networking sites.

Improving Personal Product Recommendation via Friendships’ Expansion

The trust as a social relationship captures similarity of tastes or interests in perspective. However, the existent trust information is usually very sparse, which may suppress the accuracy of our personal product recommendation algorithm via a listening and trust preference network. Based on this thinking, we experiment the typical trust inference methods to find out the most excellent friend-recommending index which is used to expand the current trust network. Experimental results demonstrate the expanded friendships via superposed random walk can indeed improve the accuracy of our personal product recommendation.

社交网络用户反馈数据的个性化推荐算法仿真

社交网络通常涉及大量的用户和内容,使得对所有用户和所有内容进行个性化推荐的计算复杂度非常高,导致推荐系统难以捕捉到用户的偏好和兴趣。因此,提出一种社交网络用户反馈数据的个性化推荐算法。通过用户时间和空间范围的相似度计算方法,结合用户之间的信任关系,准确地捕捉目标用户的个性化信息。采用全局相似度计算方法,结合用户邻居和项目邻居的相似度计算,建立全局相似度计算矩阵分解模型,采用聚类算法按用户聚类特征个性化推荐,通过构建兴趣图谱和计算用户与聚类主题之间的参与关系,计算个性化向量的相互距离和差异性指数,实现精准的个性化推荐。实验结果表明,所提方法在困惑度和平均余弦相似性指标上表现好,说明上述方法在同等的用户反馈数据个性化兴趣推荐条件下能够提供精准和与用户兴趣相似的推荐结果。