Some ideas and examples to evaluate ontologies

Ontologies are the platforms that enable the sharing and reuse of knowledge by establishing common vocabularies and semantic interpretations of terms. While ontologies may provide for reusability, sharability or both, the evaluation of their definitions and software environment is critical to the success of the final applications that reuse and share these definitions. If wrong definitions from the ontology coexist with specific knowledge formalized in the KB, the KBS may make poor or wrong conclusions. The lack of methods for evaluating ontologies in laboratories can be an obstacle to their use in companies. The paper presents a set of emerging ideas in evaluation of ontologies useful for: ontology developers in the lab, as a foundation from which to perform technical evaluations; end users of ontologies in companies, as a point of departure in the search for the best ontology for their systems; and future research, as a basis upon which to perform progressive and disciplined investigations in this area. After briefly exploring some general questions such as: why, what, when, how and where to evaluate; who evaluates; and, what to evaluate against, we focus on the definition of a set of criteria useful in the evaluation process. Finally, we use some of these criteria in the evaluation of the Bibliographic-Data ontology (T. Gruber, 1994).<<ETX>>