融合标签权值的用户模糊聚类方法研究  

Research on the User Fuzzy Clustering Method of Integration Tag Weight

作　　者：王娜[1] 葛毓彬 

机构地区：[1]郑州大学信息管理学院,河南郑州450001

出　　处：《情报理论与实践》2016年第3期140-144,共5页Information Studies:Theory & Application

基　　金：国家社会科学基金项目"泛在网络中的信息过载与信息组织模式研究"的成果;项目编号:12CTQ027

摘　　要：[目的/意义]为提高信息服务的针对性和个性化水平,提出一种融合标签权值的用户模糊聚类方法。[方法/过程]根据用户的标注行意义为构造基于各个标签的模糊相似矩阵,然后根据提出的标签权值确定流程来确定不同标签的权值,再在此基础上求得用户标注行为的模糊相似矩阵,最后通过计算确定阈值并根据阈值对用户进行聚类。[结果 /结论]实验结果表明,该聚类算法能够有效地对用户进行聚类,而且聚类准确度要优于传统模糊聚类方法。[ Purpose/significance ] For improving the pertinence of information service and the level of individuation, this paper proposes a user fuzzy clustering method involving tag weight. [ Method/process ] Given the tagging behavior of users, the paper constructs fuzzy similar matrix based on each tag. Then, the paper determines the weight of different tag based on the proposed processing of tag weight determination. On the basis of this, the paper obtains the fuzzy similar matrix of user tagging behavior. Finally, the paper calculates and determines the threshold and clusters based on threshold to the users. [ Result/conclusion ] The experiment results show that this clustering algorithm can cluster users effectively, and its accuracy is better than that of the traditional fuzzy clustering method.

关 键 词：模糊聚类 标签权值 用户聚类 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术] G350[自动化与计算机技术—计算机科学与技术]