Multidatabases (MDBs) or interoperable databases are multiple autonomous databases managed together. Unlike a distributed database management system (DDBMS), they do not have a single global conceptual schema. In many cases, database users resist integration and would like to preserve the autonomy of their databases. MDB is a natural answer to this user tendency.
The authors of this survey paper first discuss general characteristics of MDBs, then describe how existing information retrieval (IR) and videotex databases provide for interoperability. Not surprisingly, due to the relatively simple nature of the update operations of IR systems and the availability of many online IR databases, large-scale MDBs for document retrieval are already available. The paper provides examples of them. Next, the authors present two methodologies for the design of MDB systems and discuss the relationships between these approaches and the DDBMSs. Later, they give examples of the MDB capabilities of commercial relational database and IR systems and several prototypes developed in different settings. Finally, they provide pointers for future research. These include self-documenting database systems, update problems, query processing, knowledge basis, data models, and dynamic viewing.
This paper and the other papers published in the same special issue of ACM Computing Surveys on heterogeneous databases [1,2] are beneficial to a large spectrum of people, ranging from nondatabase specialists to researchers who are interested in this application of databases. The papers also unify the terms and concepts used and try to provide a basis for future research.