Federating several autonomous database systems is useful in holding together the information of several organizations that may have different interests. In Distributed Database Systems fields, Federated Database Systems are proposed to accomplish such a goal, and a variety of prototype systems with different architectures have been developed by researchers.
This paper shows that the traditional Federated Database Systems approach has a limitation in making the members of a federation to employ their database freely, then proposes a new AI approach, Message-based Federation Architecture, using Intelligent Agents to integrate several autonomous databases effectively. Specifically, this architecture adopts the knowledge-based approach and the message-based integration framework. Each member of a federation has a knowledge-base for managing the information on the local database schema and the local data offering policy. Also, the global query is processed through message exchanges between intelligent agents. Using the knowledge-base, a local intelligent agent translates data request messages from another member into local database query language and verifies whether the specified data can be offered to the request issuer or not. The main theme is that this remote query mechanism gives each member of the federation extensive liberty in managing a local database.
This paper describes metadata usage and the working mechanism of intelligent agents for submitted architecture and applies them to modeling distributed PACS. A prototype of distributed PACS is developed on the commercial relational database system with C++ programming language.
