Folkoncept: método de suporte à modelagem conceitual de ontologias a partir da aquisição de conhecimentos de folksonomias

Abstract

In this work, we present a method called Folkoncept for supporting conceptual modeling of ontologies starting with knowledge acquisition based on folksonomies. The method aims at helping actors enrolled in the development process in eliciting terms and in the modeling choice of how to represent these terms in the ontology. The objective of applying the Folkoncept method is to reduce the knowledge acquisition bottleneck through ontology learning techniques based on folksonomies. Folkoncept reaches three activities of the development process: knowledge acquisition, conceptual modeling, and evaluation, the latter being integrated into the conceptual modeling activity. With relation to the knowledge acquisition, Folkoncept deals with the retrieval, representation, and enrichment of terms (tags) coming from a folksonomy resulting from a social process of tagging performed by the actors involved in the ontology development process. In the conceptual modeling activity, Folkoncept helps the ontology designer to transform folksonomy’s tags into elements of the ontology being developed. In the ontology evaluation activity, the method helps ontology designers to validate the new elements that are suggested by the ontology learning method. In addition, the Folkoncept reduces the difficulty in using the OntoClean methodology making its use transparent to the ontology designer. Folkoncept was evaluated by means of ontology development experiments realized in a controlled environment by teams composed by ontology designers coming from the area of computing. Some teams worked with a prototype system that implements the Folkoncept. Results obtained by these teams were compared with the results from teams working without the system. The comparison was performed through metrics that show that the Folkoncept helped ontology designers to develop more descriptive ontologies with fewer errors with relation to the idealized taxonomies of OntoClean.