Tackling inconsistencies in data integration through source preferences

Dealing with inconsistencies is one the main challenges in data integration systems, where data stored in the local sources may violate integrity constraints specified at the global level. Recently, declarative approaches have been proposed to deal with such a problem. Existing declarative proposals do not take into account preference assertions specified between sources when trying to solve inconsistency. On the other hand, the designer of an integration system may often include in the specification preference rules indicating the quality of data sources. In this paper, we consider Local-As-View integration systems, and propose a method that allows one to assign formal semantics to a data integration system whose declarative specification includes information on source preferences. To the best of our knowledge, our approach is the first one to consider in a declarative way information on source quality for dealing with inconsistent data in Local-As-View integration systems. ©2004 ACM