Improving Theory: Building Robust Organizational Forms for Holonic Multiagent Systems

Recent research in distributed artificial intelligence (DAI) has shown that different organizational forms to which agents may organize themselves have beneficial and varying effects on the performance of multiagent systems (MAS) regarding scalability, agent drop-out safety and flexibility. This talk presents a concept to enlarge a spectrum of five organizational forms that has been specified with the aid of sociological research to a search space of organizational forms beyond previous frameworks. The contribution of this work is to search this space for superior forms of organization with the help of genetic algorithms in order to firstly, verify the goodness of the basic organizational forms and secondly, to establish reasons why some organizational forms behave more robust than other forms.