位置敏感的移动社交关系预测与圈子识别方法研究

项目名称： 位置敏感的移动社交关系预测与圈子识别方法研究

项目编号： No.61472299

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 黄健斌

作者单位： 西安电子科技大学

项目金额： 85万元

中文摘要： 随着移动社交网络的兴起，由于其情境感知的实时连续性和数据的时空性，使得深入分析移动用户的活动规律，预测和识别移动社交关系及强度，揭示移动社交网络的内在结构和性质，已经成为一项亟待解决的研究课题。该项研究的成果能够广泛应用于好友推荐、产品定向广告、用户活动预测等基于地理位置的服务系统中。本项目采用海量数据挖掘技术，通过深入分析用户移动社交关系数据中潜在的规律和模式，建立用户行为模型，实现基于移动轨迹的用户相似度评价指标和相似性搜索技术；通过研究位置敏感的移动社交关系预测和融合多特征的关系强度识别方法，实现移动社交网络中的好友推荐；通过融合社交链接关系、地理位置、轨迹、属性等多维度信息，分析和检测移动社交网络中的社交圈子，实现内在群组结构的自动识别。针对以上问题设计开发时间与空间效率高、多特征融合、无需用户干预、可伸缩性强、鲁棒性好的算法软件包，使其能广泛应用于各种基于位置的社交和服务系统中。

中文关键词： 移动社交网络；轨迹数据；关系预测；社交圈子

英文摘要： With the emergence of mobile social network, the real-time context-change awareness and spatiotemporal property of data of this service establish a new research subject that should be explored. This subject encompasses in-depth analysis of travel patterns of mobile device users, the recognition and prediction of friend relationships between the users of mobile social network and the social ties, and the revelation of the intrinsic structure and characteristics of mobile social networking. The research can be applied to many geolocation based systems such as friend recommendation, targeted advertising and user activities prediction etc. Built on the mining of extensive data, our project constructs the model of user behaviors and implements searching techniques relying on motion trajectories user similarity and its assessment index, provides friend recommendation through the study of context-sensitive relationship prediction and a means of discovering relationship intensity that considers several features, and analyze and examine communities in mobile social networks and realize auto internal circles detection via taking multiple factors such as links in social network, geolocation, trajectories and other attributes into account. With regard to the preceding issues, we design the algorithms and develop a software package that is of high space and time efficiency, multi-features considered, no manual effort needed, and has excellent scalability and robustness properties. We make it possible to apply this framework to various servieces and systems of geolocation based social networks.

英文关键词： Mobile Social Network;Trajectory Data;Relationship Prediction;Social Circle