Método de argumentação para resolução colaborativa de divergências na combinação de ontologias individuais

Abstract

This thesis proposes an argumentation method for collaborative solution of divergences in the merging of individual ontologies, the so-called CAMIO (Collaborative Argumentation in Merging Individual Ontologies). Such method aims to support, serve as basis, and formalize the collaborative argumentation for solving divergences deriving from the merging of ontologies individually developed. Divergences are inherent to the collaborative development of ontologies and are generally solved through negotiation processes among the participants. Commonly, during these negotiations the participants advocate their ideas based on informal argumentation, which are so called due to their tacit knowledge basis – knowledge stemming from the participant’s personal experiences and that is hard to formalize and spread, once it is specific and acquired in a specific time and space. The more argumentation primarily derives from personal experiences or individual interests of the participants, the higher the risk that the ontology deviates from the intended model. Consequently, it is harder to achieve the ontology correctness and existing errors must be reviewed and corrected in subsequent stages, increasing the ontology development cost and, if not corrected, even making its utilization not viable. The CAMIO method proposes the formalization of argumentation through the use of philosophical principles deriving from the theories of essence, identity, unity, and dependence (preconized by the OntoClean methodology) as for justifying part of the argumentation used in the negotiation among participants. The OntoClean utilization aims to facilitate reaching a consensus and reducing the distance between the intended and the specified models, once the application of the methodology helps reveal the meaning intended for the representation of concepts through the understanding of the logical consequences of the modeling ontological choices, thus assuring a consistent interpretation of the ontology elements. The evaluations of the experiments performed by using the proposed method suggest that our approach is feasible and implementable in practice.