一种改进的基于本体的Web信息抽取.pdfVIP

第36 卷 第4 期 计 算 机 工 程 2010 年2 月 Vol.36 No.4 Computer Engineering February 2010 ·软件技术与数据库· 文章编号：1000—3428(2010)04—0039—03 文献标识码：A 中图分类号：N945 一种改进的基于本体的Web 信息抽取 柳佳刚，陈 山，黄 樱 (湖南工学院计算机科学系，衡阳 421002) 摘 要：以Web 页面信息项本体定义为基础，对单个样本页面信息项路径进行启发式学习，对所有样本页面集中信息块路径进行归纳学习， 识别结构相似的信息块子树位置，以准确划定信息抽取区域，降低页面噪声。将经过噪声处理的样本页面自动解析成页面的结构本体。比 较Web 页面信息项本体和页面的结构本体，通过归纳学习算法生成抽取规则，提高Web 信息的抽准率。 关键词：信息抽取；本体；归纳学习；文档对象模型 Improved Ontology-based Web Information Extraction LIU Jia-gang, CHEN Shang, HUANG Ying (Department of Computer Science, Hunan Institute of Technology, Hengyang 421002) 【Abstract 】On basis of definition of information item ontology for Web page, the heuristic learning is conducted for information items in sole sample Web page, and inductive learning is conducted for the path of information blocks in whole sample Web collection. A method for researching the path of Document Object Model(DOM) tree is proposed. Using this method, the location of the structured-like information block subtrees can be got, the areas of information extraction can be partition accurately. The construction ontology by automatic parsing the Web page filtering the noise is created. Compared the information item ontology with the construction ontology, the information extraction rules by using reconstructing inductive learning arithmetic are generated. The precision of information extraction is improved. 【Key words 】information extraction; ontology; inductive learning; Document Object Model(DOM) 1 概述 项建立。 采用本体(Ontology)技术精确定义 Web 资源中共享的词