Structure- and Extension-Informed Taxonomy Alignment

Ontologies and concept taxonomies help software systems organize data more effectively for particular application domains. Ontologies also enable sharing and integration of data from different domains and data sources. However, ontologies from different domains are rarely identical; thus, there is need for techniques to find alignments between concepts in different ontologies and taxonomies. In this paper, we first note that alignment algorithms can be classified into two, structurally-informed and extensionally-informed. We then present a concept vector based scheme that captures structural information inherent in taxonomies to facilitate structure-based matching of concepts across taxonomies. We further note the structurally-informed concept vectors can further enable us to benefit from an available corpus of documents to implement extensionally-informed matching schemes with improved power of discriminating synonyms and homonyms of concepts.