Learning Ontologies with Relational Concept Analysis
Patricia Mateiu, Adrian Petru Groza, Cristina Nica · 2022
We propose an automated method and tool for learning ontologies from data. We rely on Relational Concept Analysis (an extension of Formal Concept Analysis) to learn concepts with their instances, relations between these concepts, and properties of these relations, such as symmetry, transitivity and reflexivity. The learned model is formalised in Description Logics. In addition, since we have access to all learned axioms, the learned model is a white box one.