基于语义标注和最小二乘神经网络的信息抽取  

Information Extraction Based on Semantic Tagging and Least Squares-Neural Network

作　　者：冯艳[1] 

机构地区：[1]山西大学计算机与信息技术学院,山西太原030006

出　　处：《科技情报开发与经济》2008年第20期138-140,共3页Sci-Tech Information Development & Economy

基　　金：国家863高技术研究发展计划资助项目(2006AA01Z142)论文

摘　　要：提出了一种基于语义标注和最小二乘神经网络信息抽取的方法,并选用教材为研究对象,以语义标注作为构建信息抽取规则的基础,以原始文本与目标模板之间的相似度作为竞争力,通过原始文本与目标模板的竞争来实现原始文本的分类和噪声信息的过滤,直接从分类的角度抽取出教材信息。This paper presents a method for information extraction based on semantic tagging and Least Squares- Neural Network, and chooses the abstract of the teaching materials as the subject investigated and the semantic tagging as the base for rule-building in information extraction, and through the competition between the primitive texts and target templates, realizes the classification of the primitive texts and the filtration of the noise information, and extracts the information abstract of the teaching materials directly from the angle of the classification.

关 键 词：信息抽取 语义标注 最小二乘神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]