On the Assessment of Concept Relevance in FCA-Based Ontology Restructuring

Along their lif-cycle, Ontologies typically undergo a number of structural operations that might deplete their structural quality. Ontology restructuring is a process of improving that quality by reconsidering the way the knowledge is spread across the class and property hierarchies. This often leads to the discovery of new abstractions whose relevance to the ontology must be assessed. We address the relevance assessment within a context where it is crucial: Our restructuring method performs a enhanced concept analysis on the initial ontology that outputs all abstractions of a predefined language. Thus, the key step is the selection of the abstractions to include into the restructured ontology. We propose a set of relevance metrics and a rule-based algorithm for combining them into a single filtering criterion. Their effectiveness is evaluated using a collection of ontologies that have been analyzed beforehand to provide ground truth.