Abstract

Comparing the communication overhead of Multi-Agent Systems (MAS) is a complex problem and it does not have a single form. Distributed architectures of MAS are assumed to have greater robustness than centralized approaches. But the corresponding cost in terms of agents interactions is not often quantified in order to fairly evaluate the relative benefits of them. The present work focuses on evaluating interaction relevance (measured through the involved concepts and produced reactions). So, in this paper, we describe the assignment of evaluation values to agents interaction of a distributed and a centralized specific MAS architectures applied to the same context-aware domain. Due to the dependant nature of the relevance of the messages, the evaluation had to be adhoc, but here we provide an example of how interesting is this alternative in order to evaluate any MAS architecture theoretically.