An Approach of Semantic Similarity Measure between Ontology Concepts Based on Multi Expression Programming

To improve accuracy of semantic similarity measure between ontology concepts, four main factors that impact on semantic similarity measure is taken into account. They are semantic distance, semantic depth, semantic coincidence and semantic density. Firstly, they were preprocessed to obtain four basic methods for calculating semantic similarity. And then Multi Expression Programming algorithm was adopted to combine and optimize the four basic methods. Thus, an approach of semantic similarity measure between ontology concepts based on Multi Expression Programming is proposed. At last, the approach is tested using dataset extracted from WordNet. The experiment result shows that the approach can be able to exclude the influence of non-key factor and enhance accuracy of semantic similarity measure.