Research On Fact-Oriented Ontology Modeling With Two-level Representation Framework

Ontology-related research has been made great achievements. Ontology-related technology has been widely used in various knowledge-based systems, and has become the infrastructure of the semantic computing systems since the philosophy term "ontology" was introduced into the field of computer and information technology in the early 1990s. Ontology modeling is one of the key activities in ontology engineering. Many challenges still need to be addressed although there is a variety of ontology modeling methods. It is necessary to study an ontology modeling method, which easy to use by non-ontological staffs, easy to reach consensus, and beneficial to improve the quality of ontology.In order to conquer the ontology modeling challenges, a methodology called FOOM, a fact-oriented ontology modeling method with two-level representation framework, has been proposed in this dissertation. Related methods and technologies of FOOM have been explored, including modeling principle of domain ontology, the management framework and formal definition of domain ontology, ontology modeling language, modeling process, ontology mapping, model merging, and semantic conflict checking patterns. The major contributions of this dissertation as following:FOOM uses fact-oriented modeling method to require and represent domain ontology, uses two-level management framework to manage domain ontology, and follows two-level modeling principle to build and maintain domain ontology. As a result, many challenges can be addressed by FOOM, including (1)fact-oriented ontology modeling approach allows domain experts to really participate in the ontology modeling process, so the quality of ontology can be fundamentally improved; (2)fact-oriented conceptual models are easy to read and understand, so the accessibility of domain ontology would be effectively improved; (3)two-level management framework is beneficial to improve the reusability and usability of domain ontology, to address the conflict between reusability and usability requirements, to build application ontology in distributed environments, to simplify the evolution process of domain ontology, and to make it easier for maintenance tasks. (4)two-level modeling principle divides the domain ontology modeling task into two levels, one is structure ontology modeling for domain level knowledge, the other is application ontology modeling for business/task level knowledge, so ontology consensus is easy to be reached.A fact-oriented ontology modeling language called FOOL has been constructed to represent domain ontology with graphical modeling primitives for viewing and discussion. Formal semantics of FOOL has been defined with first order logic for helping modelers and developers to accurately understand the underlying logic of domain ontology. Based on MOF (Meta Object Facility) specification, the ontology definition meta-model of FOOL called FOOL-ODM has been modeled in order to exchange FOOM ontology between ontology tools that support MOF, and technical staffs who familiar with UML (Unified Modeling Language) can use UML tools which support MOF to model domain ontology with FOOL-ODM. An XML Schema of FOOL called FOOM-ML has been developed to exchange FOOM ontology on Semantic Web. An abstract syntax of FOOL has been defined with EBNF (Extended Backus-Naur Form) to store FOOM ontology and to check syntax of FOOM ontology.Translating rules have been proposed for mapping FOOM ontology into OWL (Web Ontology Language) axioms and/or CL (Common Logic) expressions. By translating, FOOM ontology can be shared on Semantic Web and can be applied in knowledge management systems, and moreover it can be reasoned and checked with reasoning engines that support OWL DL or CL.FOOM model merging methods have been proposed for guiding modelers to merge models during domain ontology modeling and for designing auto-merging tools. Local semantic conflict checking patterns and corresponding instant checking algorithms have been proposed for guiding modelers to check out common semantic conflicts in FOOM ontology and for designing instant checking tools that can be used in ontology modeling and merging.