基于词嵌入语义的精准检索式构建方法  

Construction of Precise Search Queries Based on Word Embedding

作　　者：何涛[1] 王桂芳[1] 杨美妮 郭楷模 He Tao;Wang Guifang;Yang Meini;Guo Kaimo(Wuhan Documentation and Information Center,Chinese Academy of Sciences,Wuhan 430071,China;Department of Mathematics,Naval University of Engineering,Wuhan 430033,China)

机构地区：[1]中国科学院武汉文献情报中心,湖北武汉430071 [2]海军工程大学数学教研室,湖北武汉430033

出　　处：《现代情报》2018年第11期55-58,共4页Journal of Modern Information

基　　金：中国科学院"青年创新促进会"(项目编号:2016160)资助项目

摘　　要：[目的/意义]使用科技文献数据库进行文献检索时,检索式中的关键词如果不够全面,将导致检索结果查全率较低;检索式中的关键词如果一词多义,则可能向检索结果中引入无关文献,导致查准率较低。[方法/过程]针对这两类问题,本文提出使用词嵌入这一新颖的文本数据化表现形式,一方面通过语义分析对检索关键词进行扩充从而提高查全率;另一方面通过发现语义异常点来提高查准率。[结果/结论]本文将该方法应用于人工智能领域中深度学习方向上的文献检索式构建,实验结果表明该方法能在一定程度上提高检索的查全率和查准率。[Purpose/Significance]During the literature search by using the academic databases,the search query of incomplete keywords would result in a low recall ratio;besides,the search query of polysemous keywords could introduce irrelevant literature and lead to a low precision ratio still.[Method/Process]To solve these two problems,this paper presented a novel manifestation for datafication of texture,namely word embedding:on one hand,to supplement the keywords by semantic analysis so as to improve the recall ratio;on the other hand,to enhance the precision ratio by detecting the semantic outliers.[Result/Conclusion]In this paper,the method was applied to the construction of literature search queries for deep learning in the field of artificial intelligence(AI),and the experimental results suggested that this method could improve the recall ratio and precision ratio to a certain extent.

关 键 词：深度学习 词嵌入 查准率 查全率 检查式构建 

分 类 号：G252.7[文化科学—图书馆学]