Ph.D. thesis

FORT: A modular Foundational Ontological Relations Theory for representing and reasoning over the composition of tangible entities – An application to Cultural Heritage entities

Keywords: Applied Ontology, Knowledge representation and reasoning, Foundational Relations, First-order logic, Description logics, Common logic, Cultural heritage.

The slides of my Ph.D. defense presentation can be found here: DANASH-Thesis-Presentation.

Manuscript available here: DANASH-Fatme-Thèse-Manuscript

In this thesis, we address the objective of representing and modeling the composition of any
tangible entity using structural and spatial ontological relations, drawing insights from cultural
heritage. For this purpose, we propose « FORT: a Foundational Ontological Relations Theory »
within an applied ontological approach. FORT is designed with the following characteristics : (a)
modular, i.e. composed of interlinked and intralinked relation modules; (b) a meta-ontology i.e.
specifying a meta-conceptualization of top-level abstractions and using a meta-modeling language of generic modeling primitives, and (c) exclusively addressing relations and rule constraints.

To formalize FORT and illustrate its employment, we construct and adhere to an ontology
engineering methodology. This methodology addresses various specification choices for FORT,
namely expressivity and decidability, resulting in two versions: the FORT reference ontology and the FORT lightweight ontology. Furthermore, the methodology formalizes each specification, the reference, and lightweight ontologies, at multiple levels, namely theoretical and empirical. Thus, FORT is formally expressed in a First-Order Logic (FOL) formalization with a Common Logic Interchange Format (CLIF) serialization for the reference ontology, and a decidable Description Logic (DL) formalization using the SROIQ fragment with an OWL2 implementation for the lightweight ontology. Moreover, the methodology bridges the two specifications through a systematic translation from the reference FOL theory to the lightweight SROIQ fragment.

Therefore, our approach contributes in the following ways. Firstly, we propose an expressive
and well-founded language of exclusive relations and rule constraints through the FOL formalization of FORT. Secondly, we demonstrate the novelty and consistency of our proposed relations language through the CLIF serialization of FORT. Thirdly, we establish a decidable lightweight formalization of our relations language through a generic and systematic translation process, for the SROIQ formalization of FORT. Lastly, we provide this language as an OWL ontology and present different methods (direct and indirect) for its employment to support its practical use.