Detecting micro-blog user interest communities through the integration of explicit user relationship and implicit topic relations

作     者：Yu QIN Zhengtao YU Yanbing WANG Shengxiang GAO Linbin SHI 

作者机构：Institute of Information Engineering and Automation Kunming University of Science and Technology Key Laboratory of Intelligent Information Processing Kunming University of Science and Technology 

出 版 物：《Science China(Information Sciences)》 (中国科学:信息科学(英文版))

年 卷 期：2017年第60卷第9期

页      面：64-77页

核心收录：

学科分类：08[工学] 080402[工学-测试计量技术及仪器] 0804[工学-仪器科学与技术] 

基　　金：supported by National Nature Science Foundation (Grant Nos. 61175068, 61472168) Key Project of Yunnan Nature Science Foundation (Grant No. 2013FA130) Key Special Project of Yunnan Ministry of Education, Ministry of Education of Returned Overseas Students to Start Research and Fund Projects,and Science and Technology Innovation Talents Fund Projects of Ministry of Science and Technology (Grant No. 2014HE001) 

主　　题：feature mapping implicit topic guided topic model contribution of the neighboring nodes 

摘      要：In order to effectively utilize the explicit user relationship and implicit topic relations for the detection of micro-blog user interest communities, a micro-blog user interest community(MUIC) detection approach is proposed. First, through the analysis of the follow relationship between users, we have defined three types of such relationships to construct the user follow-ship network. Second, taking the semantic correlation between user tags into account, we construct the user interest feature vectors based on the concept of feature mapping to build a user tag based interest relationship network. Third, user behaviors, such as reposting, commenting, replying, and receiving comments from others, are able to provide certain guidance for the extraction of micro-blog topics. Hence, we propose to integrate the four mentioned user behaviors that are considered to provide guidance information for the traditional latent Dirichlet allocation(LDA) model. Thereby, in addition to the construction of a topic-based interest relationship network, a guided topic model can be built to extract the topics in which the user is interested. Finally, with the integration of the afore-mentioned three types of relationship network, a micro-blog user interest relationship network can be created. Meanwhile, we propose a MUIC detection algorithm based on the contribution of the neighboring nodes. The experiment result proves the effectiveness of our approach in detecting MUICs.