基于Ontology的信息抽取  

Ontology-based Information Extraction

作　　者：廖乐健[1] 曹元大[1] 李新颖[1] 

机构地区：[1]北京理工大学计算机科学工程系人工智能研究所,北京100081

出　　处：《计算机工程与应用》2002年第23期110-113,共4页Computer Engineering and Applications

摘　　要：为了提高Internet上的信息搜索效率,基于内容的搜索引擎成为迫切的用户需求。但内容信息的手工获取是一项繁重的工作。从文本或半结构化文档中自动地抽取用户关心的内容信息且表示成计算机能理解的形式是一项极具实用价值的挑战性研究。该文从知识表示与推理的角度研究了提高信息抽取智能性的途径,提出了将Ontology与模板规则相结合的技术,并针对线性模板表示的局限提出了基于二侧树结构的模板规则表示,同时实现了此表示下的假设生成———冲突消解推理。该技术成功地应用于招聘广告的信息抽取。To make information searching across Internet more efficient ,content-based search engine become s more and more efficient.However,the manual marking-up of content information is a laboring work.Automatic extraction of content information from text or semi -structured documents is a demanding and challenging technology.This paper aims at enhancing the intelligence of information extraction systems with improved knowledge representation techniques.It proposes to combine the Ontology representation with pattern rules,and to use bi -side tree as the structure of composing a pattern from pattern elements in replace of traditional sequential structure.In such representation,an information -extraction engine based on hypothesis -generation&conflict -resolution procedure is implemented.The proposed technique is successfully applied to information extraction system for job-recruitment advertisements.

关 键 词：Oontology 信息抽取 INTERNET 搜索引擎 模板规则 

分 类 号：TP393.4[自动化与计算机技术—计算机应用技术]