Computing method for semantic similarity of words based on How Net

At present,most computing methods of the semantic similarity based on How Net use the hypernym-hyponym relationship to calculate the semantic distance between words,the results of which are different from people' s subjective cognition. This paper attempts to recommend a new way,which will consider many factors affecting word similarity computing,such as sememe depth and density in How Net,to improve the method of calculating word similarity through mining and analyzing the relations between sememes. The experiment results show that the semantic similarity of words computed by the proposed way is more consistent with people's subjective cognition.