上下文感知的移动社交网络社会化挖掘与推荐技术研究

项目名称： 上下文感知的移动社交网络社会化挖掘与推荐技术研究

项目编号： No.61300192

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 牛新征

作者单位： 电子科技大学

项目金额： 23万元

中文摘要： 移动社交网络是多个有着相似兴趣个体组成的社交网络，为用户提供便捷的移动交互体验。面对移动用户动态上下文信息感知、社会属性挖掘等需求，传统互联网的社会化信息挖掘及推荐技术面临巨大挑战，目前，该问题的研究处于起步阶段。本课题将研究移动社交网社会化挖掘与推荐技术，为以社交关系为中心的个性化服务实现提供有力的支持。我们将基于模糊序列模式挖掘等理论和国外测试实验平台（JiST/SWANS），研究适用于关系隐藏较深、上下文变化快的移动社交社会化挖掘与推荐策略。研究主要包括：提出基于社会信息的轻量级能量感知路由策略，同时达到合理使用能量和保护隐私等目标；设计基于频繁特征模式树与交互关系超图的社群发现、模糊序列模式挖掘下的关键成员挖掘、基于兴趣度的关联优化推荐策略，保证用户的移动社交行为的实时性和有效性。也将构建移动社交微博应用原型和融合不同网络模式的算法评估平台等，可优化算法的快速适应性演化和调整能力。

中文关键词： 移动社交网络；个性化推荐；频繁模式挖掘；社群发现；隐私保护

英文摘要： Mobile social network, with individual users sharing similar interest of social services, can supply mobile users with the convenient interactive experience. Now, traditional socialization mining and recommendation technologies can hardly meet the unique and individual needs for dynamic context information awareness and social attributes mining. And the relevant research is only in its infancy. This project will focus on a study on the socialization mining and recommendation technology of mobile social network to provide strong support for social relationship centric individual services. Based on fuzzy sequential pattern mining model and relevant theories and foreign engineering testing platform (JiST/SWANS), feasible strategies for socialization mining and recommendation will be put forward, which can be applied to deeper hidden relationship and dynamic context information for the environment in mobile social network. The research will firstly put forward sociality-based routing strategies for lightweight energy-awareness to realize rational use of energy and privacy preserving at the same time. Three algorithms which include user community aware method of frequent pattern tree and interactive relationship hypergraph, key users discovery of fuzzy sequential pattern mining and association rule optimization based

英文关键词： mobile social network；personalized recommendation；frequent pattern mining；key member mining；privacy protection