Description

This course addresses the logical aspects of the problems of defining and
obtaining consistent information from inconsistent databases, i.e. from
databases that violate given semantic (integrity) constraints. 

The basic assumption that databases may be inconsistent departs from the
everyday practice of database management systems, where typically the
system checks the satisfaction of integrity constraints and backs out
those updates that violate them. However, present-day database
applications have to consider a variety of scenarios in which data is not

necessarily consistent. From this perspective, integrity constraints can
be seen as constraints on query answers rather than on the data or
database states.

The subject of consistent query answering in databases has received the
attention of the database and logic programming communities for the last 5
years. We summarize research carried out in the field, starting by the
seminal paper presented at the ACM Symposium on Priciples of Database
Systems (PODS 99) by Arenas, Bertossi, Chomicki.
 
The logical approaches that has been followed rely on the concepts of 
"repair" and "consistent query answer" (CQA). We
describe (a) logical specifications of the notion of repair and CQA; and
(b) methodologies for computing CQAs: query transformation, logic
programming, inference in annotated logics, and specialized algorithms.
Computational complexity issues are also discussed.