The Conceptual Basis for Mediation Services

• Authors: Gio Wiederhold;Michael Genesereth
• Venue: IEEE Expert: Intelligent Systems and Their Applications
• Year: 1997

As information systems grow, they depend increasingly on diverse, heterogeneous resources, such as databases, knowledge bases, bibliographic files, Web-based information, computational facilities, digital libraries, geographic information systems, and simulations. User typically develop and maintain these resources autonomously. While the immediate applications of these resources tend to be inventory control, payroll, production control, and the like, the data eventually become important for supporting high-level applications, such as planning and decision-making. System developers typically design decision-support applications subsequently and independently. In doing so, they must synchronize planning support with management objectives, and must rely on existing sources; rarely do they have time to build planning systems and perform the necessary data collection from scratch. Dealing with many diverse and heterogeneous sources overwhelms high-level applications by excessive emphasis on irrelevant, but crucial, details. Platform and operating system differences are commonly recognized and are getting easier to manage. Differences in interface protocols, data descriptions, abstraction levels, and precise meanings of terms being used do not yet yield well to automation. Mediators provide intermediary services, linking data resources and application programs. They provide integrated information, without the need to integrate the base data resources. In their article, the authors hope to promote conceptual coherence among the various current projects that have adopted a multilayer information architecture. Having a limited number of interface standards will eventually become essential. Before then, however, the widespread use of mediation services will require a common base of understanding and technology.