Abstract: In order to solve the problems in information retrieval when enterprise making rapid decision, this paper proposes an improved PageRank algorithm. Considering the time factor by Web page, it distributes the forward link different PageRank value based on the proportion by the similarity analysis between anchor text and Web page text. The final PageRank value is more suitable for topic-specific search engine and keeps simplicity of algorithm. Experimental result shows that the improved algorithm can effectively reduce the phenomenon of topic-drift and enhance the PageRank value of new Web page.

Key words: search engine, anchor text, Vector Space Model(VSM)

摘要： 为了解决企业快速决策时信息检索的问题，提出一种改进的PageRank算法。在考虑网页产生时间因素的同时，通过锚文本与网页主题的相似度分析按权重分配网页各正向链接PageRank值，产生的PageRank值更贴合主题搜索引擎的要求，并保持算法的简洁性。实验结果证明该改进算法能有效减少主题漂移现象，恰当提升新网页PageRank值。

关键词: 搜索引擎, 锚文本, 向量空间模型

CLC Number: 

• TP391