摘要
当前群体发现研究主要利用通联关系挖掘用户群体,未能充分利用网络中所隐含的用户社交关系,致使挖掘的群体不能真实反映用户在社会生活中的群体关系。提出一种基于用户位置信息的群体发现方法,利用序列模式挖掘算法挖掘用户位置规律序列,建立位置序列相关性度量标准,以位置规律相关性揭示用户社会活动的群体关系;结合局部相似性度量方法计算用户通信距离指数,反映用户之间的相识程度;最后采用通信距离指数对位置相关性进行加权计算用户群体相关性,再利用分裂聚类算法挖掘具有通信关系和社交关系的用户群体。实验结果表明,该方法能够有效地挖掘用户社交活动中的通信相关性和位置相关性,体现用户在现实社会活动中的群体关系。
Currently,group discovery mainly mines group structure with certain features based on communications relation- ship, which can' t take full advantage of the implicit social relations in networks and can' t truly represent the group relation- ship in the real social life. This paper proposed a group discovery method based on location information, utilized sequence pat- tern mining algorithm to mine location patterns, established location sequence similarity measurement criterion, detected social activity group relations with location pattern similarity. Then it utilized local similarity method to calculate communication simi- larity, reflected the degree of acquaintance between users. Finally, calculating group relations based on location pattern similar- ity weighted with communication similarity,it Utilized the clustering algorithm to mine groups with communication relations and social relations. Experimental result indicates ~hat the proposed method can better combines communication relations and loca- tion relations to mine group relations, embodying the group relationship of users in the reality of social activities.
出处
《计算机应用研究》
CSCD
北大核心
2013年第5期1471-1474,共4页
Application Research of Computers
基金
国家"863"计划资助项目(2011AA010604)
关键词
群体发现
通联关系
位置信息
序列模式挖掘
通信距离相似性
group discovery
communication relations
location information
sequential pattern mining
communicationr'li^tnnr.p ~im
|~rltv