文本检索的潜在语义索引法初探  

A Pilot Study of Latent Semantic Indexing for Text Retrieval

作　　者：杨梁彬[1] 

机构地区：[1]北京大学信息管理系,北京100871

出　　处：《大学图书馆学报》2003年第6期68-72,84,共6页Journal of Academic Libraries

摘　　要：传统的文本检索方式是基于提问集合和文本集合的单纯语词匹配检索,然而这并不能解决检索实践过程中存在的同义和多义问题。文章阐述了文本检索的潜在语义索引法的原理并通过实验来验证潜在语义索引可以用来解决同义和多义问题,完善检索系统的性能。The traditional method for text retrieval is based on single term matching which matches words of queries with words of documents. It can't solve the problems of synonymy and polysemy in practice. This paper presents the principle of latent semantic indexing for text retrieval, and validates that latent semantic indexing for text retrieval can be used for solving the problems of synonymy and polysemy so that to improve the performance of retrieval system.

关 键 词：文本检索 潜在语义索引法 LSI 奇异值分解 SVD 加权方法 

分 类 号：G354[文化科学—情报学]