Incoherent Ontology Debugging Based On Forgetting Strategy

Due to the different understandings of the domain knowledge,ontology developers often define some logical erroneous concepts during the process of ontology construction,ontology merging and ontology evolution.These erroneous concepts are called unsatisfiable concepts and cause the ontology to be incoherent.Ontology debugging is the main method for solving the incoherent problem.The goal is calculating the minimal unsatisfiable preserving sub-TBox(MUPS)of an unsatisfiable concept,which is the key reason of the unsatisfiability of the concept.The black-box algorithm based on "expansion-contraction" framework has become the major method for ontology debugging because of its portability and robustness.However,the black-box calls external reasoners for the satisfiability checking in the expansion stage.The checking results determine whether an axiom is inserted in the axiom set relevant to the unsatisfiable concepts.Similarly,in the contraction stage,the reasoners are also called to check the satisfiability of the axiom set.Therefore,the frequently calling leads to the inefficiency of the black-box debugging method.To solve the problem mentioned above,this study presents an optimization method based on the Forgetting strategy to reduce the size of input axioms.The proposed method can reduce the entire ontology to small sub-ontology,and perform the black-box debugging method based on the small sub-ontology to reduce the number of satisfiability checking both in the expansion stage and in the contraction stage.To achieve the above goal,this study creates forgetting rules based on the resolution method.The forgetting operations focus on the concepts and roles suitable for the Forgetting rules.Then labels are created for the forgotten concepts and roles for tracing the forgetting operations.Next,calculate the MUPS based on the candidate axiom set after forgetting operation.If axioms in the MUPS do not belong to the original ontology,then the created labels are used to restore these axioms.Then,a minimality operation is also performed for the restored set of axioms.We have also proved the availability and correctness of forgetting strategy and verified its validity in the experiments.