Staying 'on message' - mapping from A to B

eCommerce requires meaningful data exchange over a distributed network. A popular approach today typically requires a comprehensive integrated schema to which every data source or application has to map because end-to-end translation is too expensive in many situations. Apart from the difficulties in designing and extending the integrated schema, it tends to commoditise contents. This is why there are many competing standards. This paper proposes a novel framework to data exchange over a distributed network to support eCommerce. This framework allows new data sources to be added quickly and to be integrated with any other data on the fly according to the context of new business applications. To achieve this we consider how ontology-based techniques can be used to describe data source heterogeneities, and techniques to automatically reconcile mismatches and to transform data using incrementally and collectively created mapping rules. We show how this framework provides peer-to-peer message translations, thus enabling for example, the contents of a message of format A to be used to automatically fill a message of format B.