The Similarity Calculation of Concepts from Different Ontologies Based on Cosine

摘要

Currently, the calculation of similarity in ontology mapping is unilateral and complicated, because of this, it has resulted in many issues such as the calculation of similarity are not reliable. In this study, we had got a new similarity algorithm for the concepts from different ontologies: we not only considered the influence of name, instance, attribute, and structure of the concepts in designing the similarity algorithm, but also abstracted the relative feature dimensions of name, instance, attribute, and structure by comparing the concept from two ontologies. Then we constructed a vector constituted by these factors of relative feature dimensions of concept. In order to reflecte the similarity of concepts from two ontologies, Cosine similarity method had been used for calculating the angle made up by two vectors. Lastly, the method had been used to test data and compared to the conventional method, and the results of test showed that this method can effectively improve the accuracy in the calculation of similarity.