A Personalized Ranking Approach via Incorporating Users Click Link Information into PageRank Algorithm
PageRank algorithm models the important degree of a page in the Web link structure based on the hyperlink relationship without considering any other influencing factor. However, for personalized application, the users activities such as click behaviors, profile information and so on, are very important to explore the intent of the user. To develop a novel dynamical ranking algorithm, we change the original link structure via incorporating click information into the pure hyperlink web graph. Firstly, we construct a hybrid graph with user click information. Then, removing user node, we obtain a new graph with changing the edge weight of the intrinsic hyperlink graph. Based on the corrected graph, we calculate the important score of the Web page by utilizing PageRank framework. To evaluate the influence of the user click information, we present a new index to measure the changing ration of the page score. Experimental results on a real-world data set show that considering user click information as a kind of implicit relevance feedback can significantly improve personalized application.
Search Engines User Click PageRank Implicit Relevance Feedback
Fuwei Kang Xiaodong Liu Wenhuang Liu
Information Science and Technology Division Graduate School at Shenzhen, Tsinghua University Shenzhen 518055, P.R .China
国际会议
成都
英文
306-310
2010-12-17(万方平台首次上网日期,不代表论文的发表时间)