Research on Web Character Information Extraction Based on Semantic Similarity

Abstract

As for the loss of the comprehensiveness from the large amount of data when extracting information, this paper proposes a method of character information extraction based on semantic similarity algorithm to improve the comprehensiveness of the character information extraction of massive data in the network. The algorithm is put into the semantic tree to choose the synonyms of the word, and the character feature set which is extended by semantic similarity is applied to character information extraction. The results show that the recall reaches to 81.87% in the case of the accuracy rate being basically unchanged. Therefore, this method of character information extraction is obviously improving in comprehensiveness, and it can be used in network data.