摘要
针对"多义词"和"词典问题",结合文本分析和用户行为分析,提出了一种基于主题的个性化查询扩展模型。分析文本时,结合关联规则和图排序算法构建TextRank模型,脱离了对人工词典的依赖,并用此模型提取多文本主题;在用户行为分析上,使用移动时间窗口法建立用户模型,有效地捕获了当前的查询主题。查询扩展时,匹配用户主题与文本主题,选择相应的关联规则进行扩展。对结合关联规则与图排序的主题提取进行了实验,并将基于主题的查询扩展模型与其它查询扩展模型进行了比较。
To solve the ＂polysemy＂ and ＂vocabulary problem＂,a topic based personalized query expansion model is proposed.This model is combined with documents analyzing and user behaviors analyzing.In documents analyzing module,relevant rules and graphbased ranking algorithm are used to extract topics from documents without manual.In user behaviors analyzing module,time window is used to build the user model.It is an effective way to build the user model.The corresponding rules are used to expand queries by matching the document topics and user interests.In experiments,the topic extraction algorithm is demonstrated and the new query expansion model is compared with other models.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第20期4471-4475,共5页
Computer Engineering and Design
基金
上海市科普资源开发与共享信息化(二期)工程基金项目(07dz23401)
上海市重点学科建设基金项目(J50103)
关键词
查询扩展
关联规则
主题提取
用户兴趣
个性化
query expansion relevant rule topic extraction user interest personalized
