Incoherent Ontology Debugging And Repair Based On Clash Sequences

As a formalized knowledge representation,ontologies can complete information sharing and exchange in the semantic level of World Wide Web.Ontologeis are thus the key goal of semantic web and the important foundation of machine understanding.However,incoherent phenomenon often occurs in ontology building,updating,evolution,merging and so on,thereby invalidating the inference,inquiry and reuse of the ontology.The measures to solve these problems are ontology debugging and repair.The goal is to determine the cause of the incoherency for obtaining the minimal unsatisfiable preserving sub-ontology(MUPS)and then provide an effective repair method for domain experts.Ontology debugging and repair plays a key role in guaranteeing ontology correctness and maintaining ontology quality,and has theoretical research significance and practical application value.This study aims to solve the key technical problems of ontology debugging and repair,determine the cause of incoherency,and provide an effective repair method.For this purpose,an optimization strategy based on clash sequences is proposed and applied to glass-box and black-box debugging methods,dynamic ontology debugging and ontology repair.The main works of this study are outlined as follows:(1)The glass-box method based on Tableau rules and the black-box method based on the “expansion-contraction” strategy are two main methods for ontology debugging.The idea of glass-box method is to construct a complete Tableau tree with the unsatisfiable concept as the root node.However,this procedure will result in large amounts of redundant expansion because each disjunction rule can generate a new sub-Tableau tree.Black-box method is prone to selecting a large number of redundant axioms irrelevant to the unsatisfiability.This procedure consumes much time and is even difficult to solve.For solving this problem,this study proposes an optimization approach based on clash sequences to extract clash set involved in the unsatisfiability from the ontology.The basic idea is to construct clash sequences related to the basic clash models and then identify the clash set based on clash sequences.Accordingly,debugging can be rapidly performed based on the clash set because the clash set is smaller than the original ontology.Experimental results show that the proposed method provides a good outcome,especially for the large-scale ontology and the large number of unsatisfiable concepts.(2)Ontologies in real-word applications are often dynamic entities that are modified frequently.New incoherent problems naturally occur when ontologies are modified.In this case,MUPS must be recalculated when new logical errors occur.This study proposes a dynamic ontology debugging method based on clash sequences.The method first constructs clash sequences for the incoherent ontology and then reuses the results of the previous debugging for the next debugging.The goal is to avoid recalculating the MUPS over the entire ontology whenever changes occur.Subsequently,the debugging efficiency can be improved because this method can effectively reduce the amount of potential solution.Experiment results show that the proposed heuristics approach based on clash sequences can achieve a significant improvement in the dynamic environment.(3)For ontology repair,this study presents an effective repair strategy based on clash sequences.The basic idea is to identify the unsatisfiable dependent path from clash sequences.Accordingly,the repair set of root unsatisfiable concept can be easily obtained on the basis of the unsatisfiable dependent path.This strategy consists of three algorithms: obtaining clash paths from clash sequences,identifying the root unsatisfiable concepts from clash paths,and partitioning axioms relevant to the root unsatisfiable concepts.Experimental results show that the presented repair strategy can effectively obtain the repair set and avoid information loss of ontology.