Round-Trip Engineering of Ontologies for Knowledge-Based Systems

摘要

Knowledge Engineering methodologies suggest to develop knowledge-based systems by means of abstract conceptual models such as ontologies. However, they provide little support for integrating these models into the overall software architecture. As a result, moving from high-level conceptual models to a reliable and maintainable implementation is difficult and expensive. The central idea of the Software Engineering framework for knowledge-based systems presented in this paper is to amalgamate ontology construction with an object-oriented development process. The ontologies, designed as UML class diagrams with OCL constraints, are automatically mapped to structure-preserving Java classes. Meta-information on these classes, which is extracted at run-time by means of reflection, enables the reuse of generic components for knowledge acquisition, consistency checking, schema translation and knowledge sharing. The approach enables the efficient round-trip engineering of knowledge-based systems by means of smoothly integrating ontologies into the software architecture, massive tool support, compatibility with existing standards and rapid prototyping.