REBECA: Turning WordNet Databases into “Ontolexicons”

AbstractIn this paper we outline the design and present a sample of the REBECA bilingual lexical-conceptual database constructed by linking two monolingual lexical resources in which a set of lexicalized concepts of the North-American English database, the Princeton WordNet (WN.Pr) synsets, is aligned with its corresponding set of lexicalized concepts of the Brazilian Portuguese database, the Brazilian Portuguese WordNet synsets under construction, by means of the MultiNet-based interlingual schema, the concepts of which are the ones represented by the Princeton WordNet synsets. Implemented in the Protégé-OWL editor, the alignment of the two databases illustrates how wordnets can be turned into ontolexicons. At the current stage of development, the “wheeled-vehicle” conceptual domain was modeled to develop and to test REBECA’s design and contents, respectively. The collection of 205 ontological concepts worked out, i.e. REBECA´s alignment indexes, is exemplified in the “wheeled- vehicle” conceptual domain, e.g. [CAR], [RAILCAR], etc., and it was selected in the WN.Pr database, version 2.0. Future work includes the population of the database with more lexical data and other conceptual domains so that the intricacies of adding more concepts and devising the spreading or pruning the relationships between them can be properly evaluated.