智能检索系统中生成语义分词的原理及调整策略  

Principles and Tuning Strategies for Semantic Word in Intelligent Search Systems

作　　者：颜小平[1] 严长春 马顺[1] 贺隽[1] 裴非 杜寅辰 YAN Xiaoping;YAN Changchun;MA Shun;HE Jun;PEI Fei;DU Yinchen(Patent Examination Cooperation Guangdong Center of the Patent Office,CNIPA,Guangzhou 510700;Beijing Zhiguagua Technology Service Co.,Ltd.,Beijing 100196;Ucap Cloud Information Technology Co.,Ltd.,Dongguan 523008)

机构地区：[1]国家知识产权局专利局专利审查协作广东中心,广州510700 [2]北京知呱呱科技服务有限公司,北京100196 [3]开普云信息科技股份有限公司,东莞523008

出　　处：《中国发明与专利》2022年第9期42-51,共10页China Invention & Patent

基　　金：广东省重点领域研发计划项目“基于大数据智能的多层次知识检索关键技术研究及应用”(项目编号2021B0101420001)。

摘　　要：智能检索系统中的语义检索已经是专利检索中必不可少的检索方式,基于语义基准生成的语义分词对语义检索的结果有着至关重要的影响。本文从语义分词生成的原理出发,通过具体案例阐述了语义分词的选择策略和语义分词及其权重的调整策略,通过优化语义分词的调整来实现智能检索系统检索效能的提升。Semantic retrieval in intelligent retrieval systems is already an essential way of retrieval in patent retrieval,and semantic tokens generated based on semantic benchmarks have a crucial impact on the results of semantic retrieval.This paper proceeding from the principles of semantic word generation,selection strategies for semantic words and adjustment strategies for semantic words and their weights are set forth by specific cases,enabling enhanced search performance by optimizing the adjustment of semantic words.

关 键 词：语义检索 关键词 语义分词 系统切词 分词权重 

分 类 号：D923.4[政治法律—民商法学]