基于加权关联模式挖掘的越英跨语言查询扩展  

Vietnamese-English Cross Language Query Expansion Based on Weighted Association Patterns Mining

作　　者：黄名选[1,2] Huang Mingxuan(Guangxi Key Laboratory Cultivation Base of Cross-border E-commerce Intelligent Information Processing, Guangxi University of Finance and Economics, Nanning 530003 Department of Computer Science, Guangxi University of Finance and Economics, Nanning 530003)

机构地区：[1]广西跨境电商智能信息处理重点实验室培育基地(广西财经学院),南宁530003 [2]广西财经学院计算机系,南宁530003

出　　处：《情报学报》2017年第3期307-318,共12页Journal of the China Society for Scientific and Technical Information

基　　金：国家自然科学基金资助项目"面向东盟国家语言的基于完全加权正负模式挖掘的跨语言查询扩展研究"(61262028)

摘　　要：本文探讨加权关联模式挖掘在越英跨语言查询扩展中的应用。首先提出面向跨语言查询扩展的基于支持度-CPIR(Conditional Probability Increment Ratio)-兴趣度评价框架的加权关联模式挖掘算法(WARM-SCPIRICLQE)以及越英跨语言查询扩展模型,然后提出基于词间加权关联模式挖掘的越英跨语言用户相关反馈查询扩展算法。该算法将越南语查询通过机器翻译系统翻译为英文并检索英文文档,提取前列初检文档进行用户相关性判断得到初检相关文档集,采用WARM-SCPIRI-CLQE算法对该文档集挖掘加权关联规则,从规则中提取与原查询相关的扩展词实现越英跨语言查询译后扩展。以NTCIR-5 CLIR为实验语料,将本文算法与现有算法进行实验比较,实验结果表明,本文算法能提高和改善越英跨语言信息检索性能,对长查询更有效。In this paper, the application of weighted association patterns mining in Vietnamese-English Cross Language Query Expansion （CLQE） is explored. Firstly, the algorithm, WARM-SCPIRI-CLQE, of weighted association patterns mining for CLQE is proposed based on the evaluation framework, Sup- port-CPIR （Conditional probability increment ratio）-Interest, as well as the model of Vietnamese-English CLQE. And then, a Vietnamese-English CLQE algorithm of User Relevance Feedback （URF） is presented based on weighted association patterns mining between terms. The algorithm translates the original Viet- namese query into English by machine translation system so as to retrieve English documents, and extracts the top-ranked retrieved documents to carry out user relevance judgment with the aim of obtaining related document collection. The WARM-SCPIRI-CLQE algorithm is used to mine weighted association rules in the collection, and the expansion terms related to the original query are extracted from these association rules in order to accomplish Vietnamese-English cross-language post-translation query expansion. Taking NTCIR-5 CLIR as the experimental corpus, a comparison between the proposed algorithm and the existing algorithm is made, which shows that the former can improve the performance of Vietnamese-English cross language information retrieval and is more effective for long query.

关 键 词：加权关联模式 用户相关反馈 越英跨语言检索 查询扩展 

分 类 号：G354.1[文化科学—情报学] TP391.3[自动化与计算机技术—计算机应用技术]