Querying Inconsistent Databases: A Survey
=====================================================================

A quick look at real-world data unveils that data are usually not
clean. They may not satisfy the intended integrity constraints,
leading to database instances, which are called "inconsistent" w.r.t.
the constraints.

There are many ways in which the problem of inconsistency arises in
practice. For instance, integrating data from two sources, may lead to
a database with two different tuples sharing the same Primary Key value
(for example two persons with the same SSN). Another common source of
inconsistency is that DBMSs usually support only a limited class of
constraints, making it thus hard to enforce and maintain other types
of constraints.

Given the fact that database instances are inconsistent (and human
beings are for the most part consistent) a lot of work has been done
on returning "consistent" query answers over an inconsistent database.
In this talk I will try to summarize this work. Starting from the
different semantics proposed to tackle the problem, I will present the
major results in consistent query answering, mention complexity results
and also provide a brief description of some interesting techniques used.