Distributed heterogeneous information systems and schema transformation

The advantages of the object-oriented data model Cosmos as a conceptual modeling facility for distributed database architectures are highlighted. It is shown that this data model not only offers the possibility of defining highly structured abstract constructs but also provides a general-purpose programming environment for the definition of additional operations upon already existing heterogeneous data. The problem of extracting information from existing heterogeneous data representations is also addressed, and the use of a two-layered approach is suggested. In the two-layered quintessence model architecture the user can utilize highly structured objects at the upper layer, while all interschema and interlanguage transformations are performed in a transparent manner at the lower layer. A two-layered model configuration is introduced for the case of semidecentralized distributed heterogeneous information systems and it is explained how its lower layer is in a position to map heterogeneous host schemas to each other