We consider the setting of Ontology-Based Data Access (OBDA) where different and possibly inconsistent data sources are wrapped by an ontological layer. The ontology and the data (a.k.a the knowledge base) are represented within the family of ontology languages Datalog+/-. Inconsistency-tolerant approaches such as the consistent query answering (CQA) semantics use different repairing strategies making query answering unintelligible and not straightforward to the end-user. In this talk we answer the question ``How do we make query answering intelligible to the end-user when CQA is deployed?''. Logic-based argumentation offers a suitable framework to provide explanations under inconsistency in form of dialectical explanations. We show in this talk how Boolean query answering under the CQA semantics can be formalized as an argument game between two fictitious players exchanging arguments in favor and against a query Q, one who is the proponent of Q and the other is his rival. We establish completeness and soundness results and we study the dispute complexity of this argument game. The utility of such games in general is that they can serve as dialogue devices to engage the user in an explanatory dialogue about the result of the query answering. To validate the added value of dialectical explanations we present the results of an experiment that shows the impact of dialectical explanations on user's understanding of query answering under CQA.
(dans le cadre d'un recrutement post-Doc)
