Ontology-Based Measure of Semantic Similarity between Concepts

Abstract

Semantic similarity between concepts plays an important role in knowledge sharing, Web mining and semantic sense understanding. We proposed a new measure which combines the graph-based measure and information content-based measure. The measure take the condition into account that there is another ancestor concept whose information content is nearly the same with which of the nearest common ancestor (NCA). The measure constructs the concept tree by Wordnet and computes the path length of the two concepts in the concept graph, local density and the connect power of the edge, and then integrates them with edge weight and information content. The result indicates the measure perform well in the experiment.