A kind of vector space representation model based on semantic

Through the Study of semantic similarity, we proposed a vector space representation model based on semantic similarity and applied the model on text categorization with the tool of counting semantic similarity of HowNet. This model resolves the problem of semantic correlation among the characteristics which is ignored by vector space model and reduces the semantic loss of practical application systems from this problem. Finally, by the experimentation we prove that this method improves the accuracy of text categorization, and in the best result of experimentation F1 value is increased from 81.27% to 85.24%. Copyright © 2010 Binary Information Press.