基于主题与用户偏好分析的查询推荐研究

Study on Query Recommendation Based on the Analysis of Topic and User Personalization

查询日志分析作为近年来常用的查询推荐方法，常采用基于词共现的上下文来生成查询推荐。本文利用AOL日志，在词上下文分析基础之上，采用主题分析，再结合用户偏好，进行查询推荐建模，实验结果表明：采用主题分析可以显著提升查询推荐的精确度，进一步考虑用户偏好后，推荐效果又有了进一步的提升。

As a common method used in query recommendation in recent years, query log analysis often recommends queries by using the contextual information based on co-occurrence of words. On the basis of analyzing the context of words, this paper employs the topic analysis and combines the user personalization analysis to model the query recommendation by using AOL log. The final results show that the adoption of topic analysis improves the accuracy of the query recommendation significantly and the combination of user personalization analysis further improves the accuracy of the results.