Computing repairs for inconsistent databases
Sergio Greco, Ester Zumpano · 2002
The paper addresses the problem of managing inconsistencies derived from the integration of multiple autonomous information sources. We propose a general framework for computing repairs and consistent answers over inconsistent databases, i.e. databases which violate integrity constraints. A repair for a database is a minimal set of insert and delete operations which makes the database consistent. In our framework different types of rules defining general integrity constraints, repair constraints (i.e. rules defining conditions on the insertion or deletion of atoms) and prioritized constraints (i.e. rules defining priorities among updates and repairs) are considered. We propose a technique based on the rewriting of constraints into (prioritized) extended disjunctive rules with two different forms of negation (negation as failure and classical negation). The disjunctive program can be used for two different aims: compute 'repairs' for the database, and produce consistent answers, i.e. maximal set of atoms which do not violate the constraints. The technique proposed is sound and complete and more general than techniques previously proposed.