基于语义标注的信息抽取  

Information Extraction Based on Semantic Tagging

作　　者：李向阳[1] 张亚非[2] 

机构地区：[1]解放军理工大学通信工程学院,江苏南京210007 [2]解放军理工大学训练部,江苏南京210007

出　　处：《解放军理工大学学报（自然科学版）》2004年第4期39-43,共5页Journal of PLA University of Science and Technology(Natural Science Edition)

摘　　要：词性标注是引起语义缺失的根本原因 ,提出了以语义标注作为构建信息抽取规则的基础。基于语义标注的信息抽取可消除词性标注引起的 3个负面影响 ,用统一的方法来指导信息抽取过程。这种方法避开语法分析 ,具有较细的处理粒度 ,对语义规则性强的领域有一定的普遍适用性。设计了基于语义标注的 MIE(军用信息抽取 )系统 ,并对标图文本试行信息抽取。抽取结果表明 ,基于语义标注的信息抽取方式有一定的可行性和适用性。As part of speech tagging is the main reason for semantic lose, semantic tagging is suggested to be the base for rule-building in information extraction. Extraction based on semantic tagging can eliminate three negative effects caused by part of speech tagging and uses a uniform approach to guide the process of information extraction. Processing at a finer granularity level, the method has a universal appeal for information extraction in domains showing strong semantic rules. A system called MIE (a military information extraction system), which is based on semantic tagging, is designed and used to extract information from scenario texts. The result shows that the approach is feasible and has practical promise.

关 键 词：信息抽取 语义标注 有限状态自动机 军事标图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]