Automated Ontology Design Pattern Substitution

Foundational ontologies provide many benefits in modelling quality and interoperability across domains by providing generic high-level classes and object properties. The use of foundational ontologies, however, obfuscates the output of ontology verbalisation and often obscures the intended meaning through counter-intuitive or non-naturalistic representations. The ability to use foundational ontologies for modelling and convert to forms amenable to verbalisation closes the gap between high-quality modelling and real-world verbalisation applications. It also enables previously untenable ontology query generation from natural language, as modelling is constrained. This project makes use of the existing work in which alignments between design patterns of both paradigms have been identified, designing a framework and implementing algorithms for the identification and execution of candidate substitutions. We evaluate the algorithms by means of synthetic ontologies for which all candidate substitutions and their interactions are known, and thereafter assess the frequency of occurrence of pattern instances in selected real-world ontologies.