Abstract: Through analysing the relationship between sememes of Hownet two sememes of equal distance in the same tree are supposed,and public nodes play a decisive role in similarity.The farther from the root node a sememe's node is,the more detailed classification and description of information the sememe take,the similarity is bigger.This paper proposes a new method on similarity computation which is based on Hownet.The method defines a similarity computation formula among Hownet's sememes.The experiments,demonstrates that the method given in this paper has improved performance in measuring word similarity.