Since numerous ontologies are available on the Web, the requirement for merging such ontologies remains a pertinent issue in several applications. Many solutions were proposed in the literature to solve the ontology merging problem. However, these solutions just deal with the combination of two source ontologies at a time. To address the challenge of automatically merging multiple source ontologies, we propose a clustering-based approach. In our approach, named MeMO, the similarity among the source ontologies is calculated with the aim of defining the order in which they will be merged. A distinguishing point of our proposal is that we consider that better results are obtained when more similar ontologies are combined in the first place. We argue that the combination of ontologies with low level of similarity can introduce mistakes that will be carried out during the whole merging process. This argument is demonstrated in our experimental evaluation.
展开▼
