摘要： 提出了一种基于网络用户行为的搜索引擎SISI(Similar Interest,Similar access on Internet)。SISI的查询输入是一个Web文档的URL。SISI的检索模型是使用统计的方法基于网络日志中用户对文档的访问频率挖掘相关文档,充分利用了用户在相关文档判定上的潜在意识。模型的假设基础是一组兴趣相似的人访问的文档有可能相关。与传统的搜索引擎相比较,搜索引擎SISI具有系统初始化时间代价小、空间代价小等优点。同时SISI的检索优势在于可以查找那些没有显式相似内容的相关文档,尤其是在检索处理时避开了文档的类型,将文本文档和多媒体文档一视同仁。

关键词: 相关文档, Web挖掘, 网络用户行为, 搜索引擎

Abstract: This paper provides a new search engine SISI (Similar Interest, Similar access on Internet), which based on analyzing Web usage information. The input of SISI is the URL of a Web document. The retrieval model of SISI tried to mine related Web documents based on frequency of users access in Web logs, and it takes full advantage of latent human judgment in related documents. The assumption of the model is that if a group of users who have the same interest trend, those documents they have accessed are probably related. Compared with traditional search engines, SISI has low time cost on initialization and low space cost, and especially SISI can find those related documents which have least common keywords, and SISI treats text documents and multimedia documents equally without discrimination.

中图分类号: 

• TP393