An Iterative Algorithm for Ontology Mapping Capable of Using Training Data

We present a new iterative algorithm for ontology mapping where we combine standard string distance metrics with a structural similarity measure that is based on a vector representation. After all pairwise similarities between concepts have been calculated we apply well-known graph algorithms to obtain an optimal matching. Our algorithm is also capable of using existing mappings to a third ontology as training data to improve accuracy. We compare the performance of our algorithm with the performance of other alignment algorithms and show that our algorithm can compete well against the current state-of-the-art.