基于自动生成模板的Web信息抽取技术  

Study of Web Information Extraction Technology Based on Automatically Generated Template

作　　者：张彦超[1] 刘云[1] 李勇[1] 沈波[1] 

机构地区：[1]北京交通大学电子信息工程学院

出　　处：《北京交通大学学报》2009年第5期40-45,共6页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

基　　金：教育部高等学校科技创新工程重大项目(707006);通信与信息系统北京市重点实验室资助项目(35304536);北京市教育人建项目专项资助(W0810040)

摘　　要：在网络舆情分析中,人们迫切需要自动化的工具在海量信息中抽取所需要的信息,以供进一步分析利用.针对此问题,提出了基于自动生成模板的Web信息抽取方法,可以消除网页噪声,快速有效地抽取所需的网页信息.该方法通过解析器将Web文档解析成文档对象模型,根据用户需求建立抽取规则,采用自动生成模板机制,并依据模板的抽取规则对网页信息进行抽取.实验证明,该抽取方法具有较高的召回率和准确率.In online public opinion analysis, the people need for automatic tools to find the exact information among the magnanimous information sources for further analysis. This paper presented an approach based on automatically generated template to eliminate noise content and extract information from web pages efficiently. The web page was translated into Document Object Model, and then the extraction rules according to the user demand were established. Based on the above process, the templates of web pages were automatically generated. Finally, these templates extraction rules were employed to directly extract information. Experimental result shows this extraction method which has a high recall and precision is reasonable and efficient.

关 键 词：信息抽取 模板 文档对象模型 XPATH 网络舆情 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]