点击次数：

论文类型：期刊论文

发表时间：2012-01-01

发表刊物：Journal of Information and Computational Science

收录刊物：Scopus

卷号：9

期号：18

页面范围：5765-5776

ISSN号：15487741

摘要：Social tagging systems allow user to create the content, to annotate it with free-form keywords, and to share these entities with their friends. Tag, as the most striking characteristic of social tagging systems, plays an essential role for depicting items both explicitly and implicitly. However, the problems of the semantic ambiguity and the huge tag searching space hinder the efficient analysis towards to the tags. To solve the limitations, we utilize clustering based on Probabilistic Latent Semantic Analysis (PLSA) to group the semantic related tags for discovering the hidden conceptual tag clusters layer between the user and item. Different from other methods based on PLSA, our method matches the hidden concepts discovered and the tag clusters corresponding to the item cluster. The method was evaluated through a case study of whether the topic related tag clusters could be found. Additionally, four indexes are provided and the comparison to the traditional method are given to evaluate the performance. Copyright ? 2012 Binary Information Press.