摘要：Word semantic relatedness measure plays an important role in many applications of computational linguistics and artificial intelligence. In recent years the measures based on WordNet have shown its talents and attracted great concern. Many measures have been proposed to achieve the best expression possible for the degree of semantic relatedness of words. In this paper, we consider two different modified measures for computing the semantic relatedness between two words based on the path-based approach. The first measure introduces the maximum node path into the classical path-based method to compute the relatedness of words from ontology hierarchy; it mainly exploits edge-counting technique. The second one takes the definition and semantic relationships of synsets into account; it is based on the assumption that the explicit and implicit semantic relationships between synsets impose equally importance factors in the word relatedness measure. The experimental results using the proposed methods on common datasets show that our measures yields into better levels of performance compared to several classical methods. In addition, the second approach performed better than the first one.
