基于加权关联模式挖掘的越英跨语言查询扩展

发布时间：2018-09-03 09:29

【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。
[Abstract]:This paper discusses the application of weighted association pattern mining in cross-language query extension. Firstly, a weighted association pattern mining algorithm (WARM-SCPIRICLQE) for cross-language query extension based on support (Conditional Probability Increment Ratio) / interest evaluation framework and an extended cross-language query model for Yueying are proposed. Then an extended cross-language user feedback query algorithm based on weighted association pattern mining between words is proposed. The algorithm translates Vietnamese query into English and retrieves English documents through machine translation system, extracts the front-line documents for user correlation judgment, and uses WARM-SCPIRI-CLQE algorithm to mine weighted association rules for this document set. The extension words related to the original query are extracted from the rules to implement the post-translation extension of the cross-language query. Taking NTCIR-5 CLIR as the experimental corpus, the experimental results show that the proposed algorithm can improve and improve the performance of cross-language information retrieval, and is more effective for long query.
【作者单位】：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);广西财经学院计算机系;
【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)
【分类号】：G354.1;TP391.3

【相似文献】