基于PLSA的个性化Web信息检索系统  

Personalized Web Information Research System based on PLSA

作　　者：吴启明[1] 阮忠[1] 

机构地区：[1]河池学院,广西宜州546300

出　　处：《软件导刊》2008年第8期149-151,共3页Software Guide

基　　金：广西新世纪教改工程"十一五"第二批立项项目(2006A-E004);广西教育厅科研项目(200708LX322)

摘　　要：分析了用户兴趣特点及变化规律,将用户兴趣分为突出兴趣和次要兴趣两类。采用Web浏览内容挖掘和行为分析相结合的方法进行用户建模,对收藏文档采用PLSA(Probabilistic Latent Semantic Analysis)生成用户突出兴趣和潜在兴趣,对用户查询请求进行扩展以提高查全率。通过对查询结果的二次重排提高TOP-X查准率,从而实现了用户个性化的信息检索。This paper first analyzes the characteristics and variation rules of the user＇s interests, and then it categorizes the user＇s interests into two types： the stable and the secondary. To realize personal information retrieval, the paper makes some proposals： first, user modeling he done by Web browsing content mining and behavior analyzing; second, the scope of the user＇s request he expanded to improve the recall; and third, the results be re-ranked to improve the TOP-X precision after a distinction, based on the probabilistie latent semantic analysis （PLSA） of the favorite documents, is generated between the user＇ s stable interests and notential ones.

关 键 词：概率潜在语义分析 个性化 信息检索 WEB信息检索系统 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论]