关键词: 信息抽取, 语义角色, 概念图相似度, 知网, 文本理解

Abstract: Because the traditional information extraction approaches are lack of semantic information, the accuracy is not high in extraction. In order to solve the problem, this article proposed a novel method of information extraction based on semantic role and concept graph. On one hand, the process transformed the shallow semantic information into concept graphs, and accurately described the main meaning of sentences. On the other hand, the calculator computed the similarity of concept graphs so as to distinguish the different domains of information. Meanwhile, the mapping rules would be got by using semantic role for improving the quality of extraction. The experimental results show that this method of information extraction is feasible and effective.

Key words: information extraction, semantic role, similarity of concept graphs, Hownet, text understanding