Knowledge Extraction Based on Sentence Matching and Analyzing

摘要

Knowledge Extraction extracts knowledge from documents by analyzing document contents, then markups knowledge attribute and stores them into knowledge base. This paper discusses definition of knowledge extraction at first, then presents goal, key technique, application and system architecture of knowledge extraction based on sentence matching and analyzing. The system identifies new sentences in a paper, and makes an analysis of the sentences from internal structure and subject semantics. Then justify knowledge metadata by analyzing association among sentences and pragmatic of the whole paper. At last, allocates knowledge attribute tag for extracted knowledge, such as definition, historic development, characteristics, key technique, classification, application and development trend in the future. Knowledge extraction based on sentence matching and analyzing can not only justify the academic copying or scientific citation automatically, as well as make a document reviewed automatically, but also achieve a conversion on analysis from paper and chapter to sentence and paragraph, which can induce a revolution in knowledge organization and management. Therefore, the study has far-reaching theoretical significance and wide application.