Detection of corrupted schema mappings in XML data integration systems

In modern data integration scenarios, many remote data sources are located on the Web and are accessible only through forms or web services, and no guarantee is given about their stability. In these contexts the detection of corrupted mappings, as a consequence of a change in the source or in the target schema, is a key problem. A corrupted mapping fails in matching the target or the source schema, hence it is not able to transform data conforming to a schema S into data conforming to a schema T, nor it can be used for effective query reformulation. This paper describes a novel technique for maintaining schema mappings in XML data integration systems, based on a notion of mapping correctness relying on the denotational semantics of mappings.