关键词: PageRank算法, 链接分析, 向量空间模型, 搜索引擎

Abstract: In order to solve the "Topic Drift" problem of PageRank algorithm, this paper proposes an improved method combined with VSM(vector space model) technique. First, it computes PageRank value by hyperlink structure of Web page, then builds vector space model of Web page content and computes topic content similarity. Finally it sums up new PageRank value according these two values by certain weight coefficient. Contrast experiments show that improved PageRank algorithm reduces the quantity of irrelevant Web page and provides better sorting results for search engine.

Key words: PageRank algorithm, hyperlink analysis, vector space model, search engine