A Multiagent System-Approach for the Design of Information Systems in Virtual Organizations
Ulrich Wanka, Rainer Unland, Stefan Kirn · 1997
This paper presents a design approach for information systems of virtual organizations which is based on the Distributed Artificial Intelligence (DAI) concept of Multiagent Systems (MAS). Viewing information retrieval within modern organizations as an inherently distributed problem it is evaluated in how far the main DAI architectures (Distributed Problem Solving (DPS), Cooperative Distributed Problem Solving (CDPS), and Multiagent Systems) are suited to solve the specific requirements imposed by virtual organizations. It will be shown that MAS, in contrast to (C)DPS, can meet these requirements in large parts mainly because MAS can easily be designed bottom-up and, moreover, does respect the autonomy of its system members. However, unlike DPS, MAS has some weaknesses in cooperative problem solving, which is an extremely important feature, not only in virtual organizations. This paper shows how cooperative problem solving can be integrated without violating the basic characteristics of MAS.