基于兴趣主题稳定度和多维相似度的社交用户兴趣挖掘  

Social User Interest Mining Based on Stability of Interest Topic and Multidimensional Similarity

作　　者：吴树芳[1] 高梦蛟 朱杰[2] Wu Shufang

机构地区：[1]河北大学管理学院,河北保定071000 [2]中央司法警官学院信息管理系,河北保定071000

出　　处：《情报理论与实践》2022年第12期186-194,155,共10页Information Studies:Theory & Application

基　　金：河北省人文社会科学研究重大课题攻关项目“基于大数据的河北省网络治理机制研究”(项目编号:ZD202102);河北省自然科学基金项目“基于语义结构的自适应图卷积跨模态特征学习方法研究”(项目编号:F2022511001)的成果。

摘　　要：[目的/意义]针对当前兴趣挖掘研究未考虑用户兴趣稳定性和未充分整合社交关系中的兴趣资源问题,提出了利用兴趣稳定且多维相似的重要社交兴趣资源挖掘用户兴趣的方法。[方法/过程]在构建用户社交网络的基础上,利用不同时间窗口的兴趣波动幅度计算关联用户的兴趣主题稳定度,并基于网络节点信息和连接关系,计算用户间的多维相似度。融合兴趣主题稳定度与多维相似度,获取用户的重要社交关系资源,实现对用户兴趣的挖掘。[结果/结论]实验采用新浪微博数据作为测试集,与已有方法相比,新方法可以有效提高兴趣挖掘效果。[局限]仅利用单一社交平台进行研究,未从多源数据中进一步探讨用户的兴趣变化。[Purpose/significance]Recent studies on user interest mining do not take into account the stability of user interest,and do not fully integrate the resources of interest in social relationships.As a solution to these issues,a new method for interest mining is proposed using significant social interest resources with stable interest and multidimensional similarity.[Method/process]On the basis of building user’s social network,the associated user’s stability of interest topic can thus be determined by measuring interest fluctuations across multiple time windows,in addition,the degree of multidimensional similarity between users can be calculated by analyzing network node information and connection relationships.Finally,by integrating the stability of interest topics and multidimensional similarity,the significant social relationship resources of users are obtained to mine user’s interest.[Result/conclusion]Experiments adopt Sina Weibo data as the test set,and the results indicate that the new method can effectively improve the effect of interest mining compared to other methods.[Limitations]This paper utilized only a single social platform to conduct research,failing to further explore changes of user interest by leveraging information from diverse sources.

关 键 词：兴趣挖掘 社交用户 兴趣主题 稳定度 多维相似度 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]