An ontology-based approach for measuring semantic similarity between words

Abstract

The estimation of semantic similarity between words play an important role in many language related applications. In this paper, we survey most of the ontology-based approaches in order to evaluate their advantages and limitations. We also present an approach for measuring semantic similarity. As a kind of feature-based method, proposed method extracts taxonomic features from ontology, aiming to provide a high-efficient, simple and reliable semantic similarity assessment method. We evaluate and compare our approach’s results against those reported by related works under a common framework. Result demonstrated that the proposed method has higher correlation with human subjective judgment than most of existing methods.