A New Hybrid Semantic Similarity Measure Based on WordNet

Semantic similarity between words is a general issue in many applications, such as word sense disambiguation, information extraction, ontology construction and so on. Accurate measurement of semantic similarity between words is crucial. It is necessary to design accurate methods for improving the performance of the bulk of applications relying on it. The paper presents a new hybrid method based on Word Net for measuring word sense similarity. Different from related works, both information content and path have been taken into considerate. We evaluate the new measure on the data set of Rubenstein and Goodenough. Experiments show that the coefficient of our proposed measure with human judgment is 0.8817, which demonstrates that the new measure significantly outperformed than related works.