一种双层的微博用户相似度算法  

A Double-Level Microblogs User Similarity Algorithm

作　　者：熊回香[1] 叶佳鑫 Xiong Huixiang;Ye Jiaxin(School of Information Management,Central China Normal University,Wuhan 43007)

机构地区：[1]华中师范大学信息管理学院,武汉430079

出　　处：《情报杂志》2018年第6期160-166,共7页Journal of Intelligence

基　　金：国家社会科学基金项目"大众分类中标签间语义关系挖掘研究"(编号:12BTQ038)研究成果之一

摘　　要：[目的/意义]对用户在某些领域上的兴趣进行深入挖掘,计算用户在某些具体领域上的兴趣相似度,并以此为基础来得到更为准确的用户整体相似度。[方法/过程]首先利用用户标签将用户的兴趣划分为不同领域,随后以LDA主题模型为基础来对用户的微博进行分析,得出用户在不同领域上的兴趣权重,再以TF-IDF方法为基础来计算用户在不同领域上的相似度,最后结合用户在每个领域上的相似度来得出用户的整体相似度。[结果/结论]实验结果表明,该相似度计算方法能得到用户在某个具体领域上的相似度,以此为基础可以分析用户在不同领域内的关系,为后续研究打下良好的基础。[Purpose/Significance]Careful mining of users＇ interests in some areas,calculating the user＇s interest similarity in some specific areas,and using this as a basis to obtain more accurate overall similarity between users is the objective of this paper.[Method/Process]First,user＇s interests are divided into different areas based on user tags,and then the user＇s microblogs are analyzed based on the LDA topic model to obtain the user＇s interest weight in different fields,then the similarity of users in different fields is calculated based on the TF-IDF method. Finally,the overall similarity of the users is obtained based on the similarity of users in each field.[Result/Conclusion]The experimental results showthat the method of similarity calculation can get the similarity of users in a specific area,and based on this,we can analyze the relationship between users in different fields and lay a good foundation for further research.

关 键 词：微博 用户相似度 标签 LDA TF-IDF 

分 类 号：TP393.092[自动化与计算机技术—计算机应用技术]