基于Dirichlet先验贝叶斯推理的社会化标注主题聚类

Topic Clustering of Social Tagging Based on Bayesian Inference of Dirichlet Prior Distribution

本文将社会化标注系统分层分析,首层分成若干个社区,次层将主题作为主导因素,同时结合LDA(Latent Dirichlet Allocation)思想构造动态贝叶斯模型并将时间因素加入进来得到时间动态演变下的隐含社区及主题的标签集,对于为网络组织及用户有效获取提供信息资源的内在属性,提炼出社会化标注系统的有效信息。

This article analyzes the social tagging system in layers.The first layer is divided into several communities,and the second layer takes the topic as the dominant factor.In combination with the Latent Dirichlet Allocation（LDA） thinking,the article constructs the dynamic Bayesian model,and with the time factor joined in,obtains the hidden community and the topic tag set under the dynamic evolution of time.The article extracts the effective information from the social tagging system to help network organizations and users effectively obtain the inherent attributes of information resources.