Semantic Similarity Measure Based on Ontology Hierarchical Tree

The notion of finding similarity between objects is used in many domains, such as information retrieval, collaborative filtering and service matching. Objects are often modeled as sets, and the measures for comparing objects' similarity are traditionally based on set intersection. Intersection-based methods can not accurately capture the similarity when there are human intuitional relationships between entities within sets. We propose a new measure that exploiting ontology hierarchical tree and vector space model in order to produce more intuitive and semantic-based similarity. We provide experimental comparison of our measure against traditional similarity measures, and also perform a user studies. The results show how well our measure matches human intuition.