Semantic Web Ontology Construction

In Semantic Web research, ontology is extremely important. As a promising model for presenting hierarchy and semantic meaning of concepts, it is widely concerned and extensively applied to many fields in computer science and technology.The core issues of ontology construction include two points: First, how to support ontology construction, the second is how to ensure that the ontology constructed is correct. For the first issue, the ontology construction tool OntoMAN is developed to support building ontology manually, and ontology automatically building is discussed based on metadata of knowledge resources. For the second one, the approach for identifying Minimal Unsatisfiability-preserving Sub-TBoxes (MUPS) based on discrimination rules is proposed to ensure correct construction.The ontology construction tool OntoMAN not only achieves the basic management of ontology entity, but also integrates the mechanism of analyzing the unsatisfiable class, which help to build ontology.An automatic ontology construction approach based on metadata of knowledge resources is proposed by integrating clustering algorithm and Open Directory Project (ODP). Ontology concepts are generated by clustering documents based on their metadata, and concept hierarchy is formed based on the hierarchy of mapped concepts in ODP. Adaptive ontology learning is conducted to update the ontology and enrich ontology concepts based on the changes of cohesion and correlation of clusters. Experimental results demonstrate the validity of the approach.The premise of repairing unsatisfiable class C is to acquire MUPS of C. Consider the transitivity of unsatisfiability, MUPS are classified into three types, fully dependent on C (MUPSf), transitively dependent on C (MUPSt) and uncertainly dependent on C (MUPSu). Based on the classification modelers and domain experts can have a clear direction on the repair work. A MUPS identification approach is proposed based on two equivalent transformations, that is, axiom splitting and ontology reduction, and three discrimination rules. The experiments show the significance of the classification for the evaluation of the quality of ontologies and for the repair work.