An approach to measuring semantic similarity and relatedness between concepts in an ontology
Yunzhi Jin, Hua Zhou, Hongji Yang, Yong Jun Shen, Zhongwen Xie, Yong Hai Yu, Feilu Hang · 2017
Semantic similarity and relatedness are applied more and more extensively in many fields, such as in Artificial Intelligence, Semantic Web and Knowledge Management. In this paper, we propose a comprehensive metric of similarity, a method of relatedness measure and a comprehensive degree measure that combines semantic similarity and relatedness between two concepts. Then we compare the proposed metrics with other classic similarity/relatedness metrics based on a benchmark data set and WordNet 3.0. The experimental results demonstrate that the proposed approach outperforms other classical computational approaches. The comprehensive degree metric gives the highest correlation value which is 0.913, and the semantic similarity measure obtains the second-highest correlation value which is 0.823 with a benchmark based on human similarity judgements.