The evaluation of Multi-Agent Systems (MAS) is a complex problem and it does not have a single form. Much effort has been spent on suggesting and implementing new architectures of MAS. Often these new architectures are not even compared to any other existing architectures in order to evaluate their relative benefits. The present work focuses on interactions, the most important characteristic of any complex software as autonomous agents according to , as a problematic of evaluation. So, in this paper, we describe the assignment of evaluation values to Agents interaction in a specific MAS architecture. This evaluation is based on the weight of the messages brought by an interaction.