Algorithm Improvement of Vocabulary Semantic Similarity with HowNet

Abstract

The vocabulary semantic similarity calculation is one of the key technologies of natural language processing, and it has been widely used in many fields such as information extraction and text classification. So how to effectively acquire the information relationships is the core issue in the study of data mining and knowledge organization.This research puts forward the concepts of semantic correlation degree and the same semantic ratio according to the semantic system structure of HowNet, and designs a computation method of vocabulary semantic similarity based on semantic distance and the degree of semantic association. The method not only considers the effect of sememes' relation to semantic similarity, but also takes into account the lexical semantic correlation degree to semantic similarity. Experimental results prove that the improved algorithm effectively improved the veracity and accuracy of semantic similarity computation method, which makes the consequence satisfy people's subjective cognition and more reasonable.