Web用户兴趣模型在个性化信息服务中有着非常重要的作用。本文利用社会化标签的独特优势,针对传统社会化标签聚类方法的局限性,提出了一种基于密度聚类的Web用户兴趣建模方法。首先建立基于社会化标签的向量空间模型,并将社会化标签表示为Web资源及其权重的形式,以此为基础利用DBSCAN算法对其进行聚类,进而依据所有Web用户的标注行为以每个聚类为中介计算特定Web用户对Web资源的兴趣度来构建Web用户兴趣模型。实验结果表明了该方法的优越性。
A Web user's interest model plays a very important role in the personalized information service.Using the tag's unique advantage,this paper proposes an approach of the Web user interest modeling based on density-based clustering algorithm against the limitations of the traditional tag clustering methods.Firstly,a vector space model based on the tags is established,and the tags are described in the form of web resources and their weights,which can be used to cluster the tags based on the DBSCAN algorithm.Then each clustering is viewed as an intermediary to account the special Web user's interest for the Web resource,according to the tagging behavior of all Web users.The experimental results indicate the superiority of this approach.