中文问答系统中机构名的处理  

Processing organization name in Chinese question answering system.

作　　者：韦向峰[1] 张全[1] 吴晨[1] 袁毅[1] 

机构地区：[1]中国科学院声学研究所,北京100080

出　　处：《计算机工程与应用》2008年第7期196-198,205,共4页Computer Engineering and Applications

基　　金：国家重点基础研究发展规划(973)(the National Grand Fundamental Research 973 Program of China under Grant No.2004CB318104);中国科学院声学研究所“所长择优基金”(No.GS13SJJ04)

摘　　要：探讨问句中机构名的处理,并服务于中文问答系统。采用概念层次网络理论的语义概念分析方法分析问句,去掉与机构名无关的概念词语,得到候选机构名。对机构名全称库按字索引,在库中搜索出与候选机构名匹配的机构名全称并按拟合权值排序。实验结果表明该方法识别机构名的正确率达到90.6%,支持对机构名简称的处理。To discuss the processing organization name and improving the performance of Chinese question answering system,an approach of semantic conceptual analysis based on hierarchical networks of concepts theory is adopted to remove the words which are irrelative with organization name and to get the remain as organization name candidates.After sorting the database of the full names of organizations ,the matched full names was selected out from the database and sorted according to the sameness as the candidates.The experimental result shows that the accurate rate is up to 90.6%.The approach can process the abbreviation of organization name.

关 键 词：问答系统 机构名 概念层次网络 按字索引 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]