Modularity in ontology change

Abstract

Nowadays, ontologies described in languages such as RDF and OWL are widely used for representing the conceptualization of a domain that can be processed by computers. Reflecting their application, ontologies may need to be updated frequently. The large amount of unstructured information encoded in ontologies makes this already complicated task even harder. We study how ontology modularisation techniques can be applied to the existing algorithms of ontology revision, debugging and evolution, since only few of them make use of modularity strategies. We focus particularly in the atomic decomposition of the ontologies and modules obtained from such structure, motivated by its logical properties and computational efficiency results. Even though there are already many studies on this concept, not many of these investigate the relations between atomic decomposition and ontology revision. Thus, we propose a study of the relations between modularisation techniques based on atomic decomposition and algorithms for ontology revision. The investigation will focus on modifying the ontology revision procedures to include modularisation as an intermediate step, and analyse the effects of the different module notions in the running time of the algorithms. On the other hand, we use the atomic decomposition of the ontology to identify the structural changes induced by the operations analysed. The study of the dynamics of the ontology structure should allow us to both optimize the current algorithms of ontology change and define new operators over the modules of an ontology. We also investigate the theoretical aspects of these alternative operators, identifying their logical properties and characterizing them with postulates. Ultimately, we compare the empirical and theoretical features of the modular operators with the existing approaches for ontology change. (AU)