基于用户行为分析的搜索引擎模型

介绍了目前搜索引擎技术发展的进展和研究热点.给出一种从用户点击的反馈信息中提取用户智能的搜索引擎算法.此算法首先给页面设置不同的内容相关度,再实测网页在不同内容相关度下的用户点击率,根据这组相关度和相应的用户点击率数据,可以由差商法或拟合直线法求得最佳的网页内容相关度,这样就可求出最终的索引相关度.实验表明该算法对提高搜索的稳定性和搜索精度有明显作用.

Page Ranking Program Based on Users' Feedback

This article introuduces a new search engine page-ranking program which can distill the users' intelligence from the users' feedback information. This program sets different relativities to a page in the first place, then measures the page's click probability of the users' feedback. Through the data of the relativities and the click probability of the users' feedback, the best page relativity by the method of difference quotient or least square linear approximation can be found. Thus, the final index relativity and the page-rank are worked out. Since the best page relativity and the page-rank are based on the users' understanding of the page's contents, they show the true relativity of the keyword and the page.