基于网络用户行为的相关页面挖掘模型

文章提出了一种基于网络用户行为的相关页面挖掘模型。模型采用统计的方法对proxy日志进行挖掘。模型的输入是一个WEB页面，输出是一组与之相关的页面。模型的假设基础是一组兴趣相似的人访问的页面有可能相关。模型从用户群中找出对输入页面感兴趣的用户，通过聚类从这些用户中找出一类具有相似兴趣背景且对输入页面最感兴趣的用户，综合这类用户感兴趣的页面，从中挖掘出与输入页面相关的页面。该模型与目前流行的相关页面检索算法的最大区别在于分析的对象是网络用户行为，因为模型认为页面是否相关的最终判定者应该是用户，通过分析网络用户行为能够更好地挖掘用户在页面相关判定上的潜在意识。用户对页面的兴趣度基于用户对页面的访问频率而定义。实验表明，该模型是可行的。该模型可用于改进传统的IR，提供相关反馈和查询扩展，使其更加适应Internet检索。该模型也可用于内容安全方面的相关主题预测。

A Model for Mining Related Pages Based on Usage Information in WWW

This paper provided a model for mining related Web pages based on Web usage information. The input to the model is the URL of a page,and the output is a set of related Web pages. The assumption of the model is that if a group of users who have the same interest trend,those pages they have visited are probably related. The model first to find a group of users who have the same interest trend and are most interested in the input page,then to collect those pages in which those users are interested,and at last to mine related pages from those collected pages. Contrast to other popular approaches in searching related Web pages,the model relies on Web usage information. Model claim that human should be the authority to judge what are really related Web pages,and Web usage information can help to make good use of latent human judgment in related pages. Experiments show the model is practicable. The model can be used in relevance feedback in traditional IR,and can also be used to predict related topics in Web content security.