Measuring semantic similarity using wordnet-based context vectors

Semantic relatedness between words or concepts is a fundamental problem in many applications of computational linguistics and artificial intelligence. In this paper, a new measure based on the semantic ontology database WordNet is proposed which combines gloss information of concepts with semantic relationships, and organizes concepts as high- dimensional vectors. Other relatedness measures are compared and an experimental evaluation against several benchmark sets of human similarity ratings is presented. The context vector measure is shown to have one of the best performances.