Conditional Mapping in Data Mediation

A serious problem facing many organizations today is the need to share information among systems that have been developed separately. Conflicts in the structure and semantics of these disparate data sources create major obstacles to effective sharing. One approach to solving this problem is data mediation. The data mediation approach relies on a common ontology that can be used to describe the structure and semantics of each of the systems that wish to participate in the information sharing. In our implementation, the common ontology is expressed as a shared conceptual schema, and we use XML to map data elements of real databases onto the conceptual schema. We have found that these mappings sometimes depend on the value of a data element or the value of some other element of the same tuple. Interestingly, the mapping may also depend on the values of data elements in the (unpopulated) conceptual schema.