Discovering Semantically Similar Associations (SeSA) for Complex Mappings between Conceptual Models

There is an increasing demand for discovering meaningful relationships, i.e., mappings, between conceptual models for interoperability. Current solutions have been focusing on the discovery of correspondences between elements in different conceptual models. However, a complex mapping associating a structure connecting a set of elements in one conceptual model with a structure connecting a set of elements in another conceptual model is required in many cases. In this paper, we propose a novel technique for discovering semantically similar associations (SeSA) for constructing complex mappings. Given a pair of conceptual models, we create a mapping graph by taking the cross product of the two conceptual model graphs. Each edge in the mapping graph is assigned a weight based on the semantic similarity of the two elements encoded by the edge. We then turn the problem of discovering semantically similar associations (SeSA) into the problem of finding shortest paths in the mapping graph. We experiment different combinations of values for element similarities according to the semantic types of the elements. By choosing the set of values that have the best performance on controlled mapping cases, we apply the algorithm on test conceptual models drawn from a variety of applications. The experimental results show that the proposed technique is effective in discovering semantically similar associations (SeSA).