Research On Graph-Based Inconsistence Handling Of DL-Lite Ontology

The rapid development of the Internet brings massive data. Whereas, semi-structured and non-structured data can hardly fit for automatic processing. The Semantic Web provides a solution for machine-readable data management. This also promotes human-computer interaction. As an important component of Semantic Web, Web Ontology Language (OWL) provides logic foundation for Semantic Web. Among the families of Description Languages which constitute the logical underpinning of OWL, DL-Lite family are able to capture the main expressivity of Entity Relationship (E-R) model and Unified Modeling Language (UML) class diagrams. So far, DL-lite has attracted many attentions from researchers, developers and users.In a practical application, the development and evolution of ontologies are complex and error-prone. Thus, logical conflicts can easily occur in ontologies and lead to logical inconsistencies. Reasoning with inconsistent ontologies will then generate unacceptable conclusions. Therefore, inconsistency of ontology is a critical issue that has to be tackled. There exist two methods to tackle this problem, one is to locate and to eliminate the cause of inconsistency; the other one is to obtain meaningful conclusions by non-standard reasoning on inconsistent ontologies. We focus on handling inconsistency in DL-Lite. The main contributions of this work are the following:(1) With respect to ontology debugging approaches. A novel ontology debugging approach based on graph is proposed for DL-Lite. In the process of ontology revision, a DL-Lite ontology is encoded to a graph. In this way all the minimal incoherence-preserving subtenninology (MIPS) of an ontology can be calculated by backtracking some pairs of nodes in the graph. The experimental results show that our debugging approach is efficient and outperforms the existing systems.(2) With respect to ontology revision approaches. We propose a graph-based ontology revision approach. In the process of ontology revision, the notion of revision state is employed to divide the terminologies of an ontology into two disjoint sets:the set of wanted axioms and the set of rebuttable axioms. We further define a revision operator based on the revision state. Afterwards, two revision algorithms are proposed to instantiate the revision operator:one is based on scoring function, and the other one is based on hitting set tree. We implement these algorithms and conduct experiments of ontology revision. The experimental results show that the algorithm based on scoring function is more efficient than that on hitting set tree.(3) With respect to inconsistency tolerance. Two inconsistency-tolerant semantics are newly defined for DL-Lite. In contrast to the classical inconsistency-tolerant semantics, the newly defined inconsistency-tolerant semantics can reserve the expressivity of the primitive semantics and do not need to calculate the closure of ABox with respect to the corresponding TBox. The experimental results show that the approach for ABox repair based on the newly semantics improves the efficiency of reasoning.(4) With respect to query answering under inconsistency-tolerance semantics. A graph-based approach for query answering under inconsistency-tolerant semantics is proposed. The query answering approach focus on the IPAR-semantic where the given ontology and the target query are both transformed into graphs by different rules and stored into graph database. The experimental results show that new approach outperforms the existing methods with respect to both efficiency and scalability.