Query Rewriting for Inconsistent DL-Lite Ontologies

Abstract

Abstract. In this paper we study the problem of obtaining meaningful answers to queries posed over inconsistent DL-Lite ontologies. We consider different variants of inconsistency-tolerant semantics and show that for some of such variants answering unions of conjunctive queries (UCQs) is first-order (FOL) rewritable, i.e., it can be reduced to standard evaluation of a FOL/SQL query over a database. Since FOL-rewritability of query answering for UCQs over consistent ontologies under first-order logic semantics is one of the distinguishing features of DL-Lite, in this paper we actually identify some settings in which such property is preserved also under inconsistency-tolerant semantics. We therefore show that in such settings inconsistency-tolerant query answering has the same computational complexity of standard query answering and that it can rely on well-established relational database technology, as under standard DL semantics.

Evaluation of Techniques for Inconsistency Handling in OWL 2 QL Ontologies

“…We now briefly recall the query rewriting technique proposed in [6]. Such a technique computes a first-order query Q starting from a union of conjunctive queries Q and a DL-Lite A TBox T .…”

“…The query Q is a perfect reformulation of Q with respect to T under the IAR semantics, i.e., Q is such that, for every ABox A, the answers to Q over T , A under the IAR semantics correspond to to the answers to Q computed over the ABox A only. Due to space limits, here we just report the main definitions of the query rewriting technique: we refer the reader to [6] for more details on the method. The first definition that we give can be used to establish whether a certain atom is consistent with the TBox axioms.…”

“…In particular, in [5] it was proved that computing the ABox repair of a DL-Lite A ontology under the IAR semantics is a tractable problem. Then, in [6] a technique for query answering under IAR-semantics in DL-Lite A is presented: instead of modifying the ABox, this method is based on computing a rewriting Q of the initial query Q and then evaluating the query Q with respect to the original ABox.…”

“…We argue that the results of [5,6] are potentially very important from the practical viewpoint, for the following reasons: (i) they are based on formally grounded notions of inconsistency-tolerant semantics; (ii) they identify (to the best of our knowledge) the first inconsistency-tolerant semantics in DLs for which query answering is tractable. So, based on such results, in principle it might be possible to define practical algorithms for handling inconsistency in OWL 2 QL.…”

“…In Section 2, we present a detailed algorithm for computing IAR repairs in DL-Lite A . In Section 3, we briefly recall the algorithm presented in [6] for consistent query rewriting under IAR semantics in DL-Lite A . In Section 4 we present the QuID system and report on the experimental evaluation we have conducted with QuID.…”