Applying a Lexical Similarity Measure to Compare Portuguese Term Collections

Marcirio Silveira ChavesVera Lúcia Strube de Lima

The number of ontologies publicly available and accessible through the web has increased in the last years, so that the task of finding similar terms1 among these structures becomes mandatory. We depict the application and the evaluation of a new similarity measure for comparing Portuguese Ontological Structures (OSs) called Lexical Similarity (LS). This paper describes contributions to the study and application of mapping between terms present in multidomain OSs. In order to approach this mapping we combine preliminar similarity measures and heuristics. Our measure uses a stemmer, it is established upon String Matching (SM) proposed in [1] and it was evaluated by means of a comparison to human evaluation. Finally, we concentrate on the application of LS measure to terms belonging to same domain thesauri and discuss the results obtained.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: