Mediation queries adaptation after the removal of a data source

Abstract EN

A broad variety of data is available in distinct heterogeneous sources, stored under different formats : database formats (in relational and object-oriented models), document formats (SGML / XML), browser formats (HTML), message formats, etc.

The integration of such data is increasingly important for modern information systems applications such as data warehousing, data mining, and web applications.

This is realized by providing a uniform view of data sources (called mediation schema or global schema) and defining a set of queries (called mediation queries or mediation mappings) which define objects of the mediation schema.

One of the important problems that merit consideration is the impact of schema evolution on mediation queries.

Mappings left inconsistent by a schema change have to be detected and updated.

In particular, one source may be removed from the system because it provides always obsolete information or because it is unavailable.

In this case it is necessary to update the inconsistent mappings.

In this paper, we study the removal of a source from an integration system and show how to correctly update the mappings between the mediation schema and the distributed sources after this change, in the context of the global-as-view approach (each relation of the global schema is expressed as a view on the data source).