Current algorithms can not to be used optimally in automatic and semi- automatic ontology matching tasks as those envisioned by the Semantic Web community, mainly because of the inherent dependency between particular algorithms and ontology properties such as size, representation language or underlying graph structure, and because of performance and scalability limitations. In order to cope with the first problem we designed a generic matching framework which exploits the valuable ideas embedded in current matching approaches, but in the same time accounts for their limitations—for specific input ontologies it optimizes the matching results by automatically eliminating unsuitable candidate matching methods.
