移动社交网络环境下的真实社会关系估计  

REAL-WORLD SOCIAL RELATIONSHIP ESTIMATION IN MOBILE SOCIAL NETWORK ENVIRONMENT

作　　者：吕明琪[1] 王琦晖[1] 胡克用[1] 

机构地区：[1]杭州师范大学,浙江杭州310012

出　　处：《计算机应用与软件》2015年第1期51-54,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(61202282);浙江省自然科学基金项目(LY12F02046)

摘　　要：移动设备的普及促进了移动社会网络的形成。移动社会网络基于用户的物理特征为其提供虚拟交互,而用户在真实世界中的社会关系知识可有效地用于改善移动社会网络服务质量。提出一种基于用户物理特征(包括由GPS坐标代表的位置特征和由蓝牙检测代表的邻近特征)估计其真实社会关系的方法。该方法首先基于用户的GPS轨迹挖掘语义化访问地点(如家、工作地点等),然后结合语义化访问地点和邻近特征估计用户间的社会关系类型。实验结果表明该方法可准确估计三种类型的真实社会关系(包括家人、同事和朋友)。The popularisation of mobile devices promotes the formation of mobile social networks （MSN）, which provides virtual interactions for users based on their own physical features, while users＇ social relationships knowledge in real-world can be effectively employed to improve the service quality of MSN. In this paper, we propose an approach for estimating real social relationship of users on the basis of their physical features including the location feature and proximity feature, which are respectively represented by GPS coordinates and Bluetooth detections in mobile environment. The approach first mines the places with semantic meanings （ e. g. home, workplace, etc. ） according to users＇ GPS trajectories, and then estimates social relationships types between the users by integrating both the semantic visited places and the proximity feature. Experimental results show that our approach can accurately estimate three types of real-world social relationships （ i. e. family members, colleagues and friends）.

关 键 词：移动社会网络 时空数据挖掘 上下文感知计算 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]