Probabilistic Answers over Inconsistent Knowledge Bases

Consistent query answering is a generally accepted approach for querying inconsistent knowledge bases. A consistent answer to a query is a tuple entailed by every repair, where a repair is a consistent database that "minimally"differs from the original (possibly inconsistent) one. This is a somewhat coarse-grained classification of tuples into consistent and non-consistent does not provide much information about the non-consistent tuples (e.g., a tuple entailed by 99 out of 100 repairs might be considered "almost consistent"). To overcome this limitation, we propose a probabilistic approach to querying inconsistent knowledge bases, which provides more informative query answers by associating a degree of consistency with each query answer by associating a probability to each repair, depending on the changes needed to obtain it.