基于粗糙集的个性化Web搜索系统

Personalized Web Search System Based on Rough Sets

本文提出了一种基于粗糙集理论的个性化Web搜索系统。用户偏好文件中对关键字进行分组以表示用户兴趣类别。利用粗糙集理论处理自然语言的内在含糊性,根据用户偏好文件对查询条件进行扩展。搜索组件使用扩展后的查询条件搜索相关信息。为了进一步排除不相关信息,排序组件计算查询条件和搜索结果之间的相似程度,根据计算值对搜索结果进行排序。与传统搜索引擎进行了比较,实验结果表明,该系统有效地提高了搜索结果的精度,满足了用户的个性化需求。

In this paper, a novel rough set based approach is proposed to create a personalized Web search system. Firstly, user profiles which consist of categories of user＇s interests by grouping related keywords are designed. Rough set theory is used to deal with inherent ambiguities of natural language and refine query according to user profiles. Then refined query is submitted to search component. To further filter out irrelevant documents for the user, retrieved resuits are re-ranked according to rough similarity measures between refined query and documents by ranking component. Experiments compared with traditional search engine are presented and experimental results indicate the precision of Web retrieval is greatly improved and system are suitable for individual usage.