| On the envisaged Semantic Web, ontologies are used to make the semantics of Web contents explicit. This enables software agents to automatically exchange information and to be assured of the meaning of the data they have exchanged. The successful operations of ontology-based intelligent applications rely on the scalable and efficient ontology management frameworks that support collaborative ontology development, evolution and reuse. The goal of this dissertation is to develop such a framework and investigate related research problems.;In this dissertation, we introduced the Temporal Description Logic SHIQ(T) as the logical foundation for reasoning about changes during ontology evolution. We conducted in depth study of automated reasoning services in SHIQ(T) and related tableau-based decision procedures. We formally proved the termination, soundness and completeness of the proposed tableau reasoning algorithms.;We proposed a set of metrics for measuring the semantic implications of changes during ontology evolution. Our metrics focus on the changes of classes and their referencing axioms and annotations in an evolving OWL ontology.;We reviewed three major large-scale ontology engineering projects and existing persistent ontology management systems and identified that insufficient support for ontology evolution, ontology modularization and fine-grained access control is their major deficiency.;We used an evolutionary log to track the history of each axiom or annotation that appears in an evolving ontology. We introduced the formal model for our ontology evolution framework and presented related algorithms. We further extended our ontology management framework to support managing locality-based ontology modules and their evolution.;We developed an axiom-centric persistent storage model for the large OWL ontologies using a relational database management system as the back-end. Our implementation supports ontology evolution, ontology modularization and fine-grained access control to the ontological axioms and annotations and provides practical reasoning capability as well as high performance querying services. We evaluated the loading time (parsing, classification, importing, module extraction), query response time, module and subsumption maintenance costs for our framework using a set of benchmark ontologies. The experimental results clearly demonstrated the scalability and efficiency of our ontology management framework. |
