|
Abstract : |
We show that the conceptual parallelism underlying multi-agent systems (MAS) and the parallelism as thought in the field of distributed systems in computer science are not the image one of each other. One should distinguish between the two notions of coexisting Agents and distributed processes in order to make them converge afterwards. Otherwise, the risk is to reduce the parallelism of the MAS to the one of the target architecture and to loose the expressiveness and the richness of the parallelism of the agents. We focus on the time conceptualization: we differentiate crude coordination of processes by synchronization from Agent temporality seen as a corner stone of autonomy. We gain insights as well into parallelism than into MAS methodology., |
