基于Agent的智能化元搜索引擎个性化机制  

Personalization mechanism in an intelligent agent-based metasearch engine

作　　者：李青山[1,2,3] 王俊[1,3] 褚华[1,2] 季陶然 

机构地区：[1]西安电子科技大学软件工程研究所,西安710071 [2]西安电子科技大学软件学院,西安710071 [3]西安电子科技大学计算机学院,西安710071

出　　处：《中国科学：信息科学》2015年第5期605-622,1-6,共18页Scientia Sinica(Informationis)

基　　金：国家自然科学基金(批准号:61373045;61173026;61202039);中央高校科研业务基金(批准号:K5051223008;BDY221411);十二五国防预研项目(批准号:513***103E);国家高技术研究发展计划(863计划)(批准号:2012AA02A603)资助

摘　　要：大数据环境下,信息量过载,人们需要精准、智能的检索工具.本文研究了基于Agent的智能元搜索引擎中的个性化机制,准确地理解用户的搜索意图,有效地提高了信息检索的服务质量.文中着重研究基于Agent的智能元搜索引擎个性化方法及功能实现所需的相关理论与技术,给出了查询语句分析与查询兴趣挖掘及成员搜索引擎调度过程,设计了基于动态学习的复杂查询识别机制,基于动态更新的用户兴趣概貌模型的检索兴趣挖掘机制,以及基于概念格与日志分析的搜索引擎评估调度策略机制.最后,针对复杂查询语句识别、搜索引擎调度策略效果及检索结果相关性的测试结果表明,本文提出的基于Agent的智能元搜索引擎个性化机制,可较为准确地识别出复杂的查询语句并进行预处理,高效学习用户的查询兴趣,达到明显提高检索结果相关程度的目的,并智能化地调度成员搜索引擎,为提高用户信息检索效率提供充分支持,从而提高用户的检索体验.Information overload is a common issue in ＂big data＂ environments, because of which such environments require precise and intelligent retrieval tools. In this paper, we propose the design and implementation of a personalization function for an intelligent, agent-based meta-search engine that improves information retrieval efficiency in ＂big data＂ environments. We focus on individualized methods, along with the relevant theoretical and technological requirements for the implementation of a personalization function in an intelligent, agent-based meta-search engine, provide query analyses, and describe the interest mining and scheduling processes of the search engine. We then detail the recognition mechanism of complex inquires based on a dynamic learning process, the interest mining mechanism based on the general view model of user interest, which involves dynamic updates, and the strategy for search engine scheduling based on a concept lattice and user logs. Experiments showed that the personalization function for intelligent, agent-based meta search engines proposed here accurately identified complex queries, effectively learned users＇ search interests, improved the relevance of the search results,and intelligently scheduled member search engine. It also helped improve the efficiency of information retrieval by users, thus enhancing their searching experiences.

关 键 词：信息过载 元搜索引擎 个性化机制 AGENT 检索相关程度 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]