Multi-expert systems are well suited to the solution of problems involving several specialists acting in a focused manner. They are less useful in cases where one faces complex systems in which many specialists must cooperate during long periods of time (like several years). In the latter case, systems based upon independent agents seem much more interesting, in part because they allow to develop more modular systems which can evolve dynamically. However, developing such open systems brings a number of difficulties such as consistency of indpendent representations, reliability of communication and cooperation processes, or lack of development tools for building applications. We worked on the various problems and can now offer some partial solutions.
Return to main page |