 |
 |
Self-interest is one of the key characteristics of an agent based system. When multiple software agents with their individual objectives are deployed in a multi-agent platform, their possibly conflicting objectives may disrupt the overall cooperation requirements of a shared environment. As a consequence, interactions of individual agents should be coordinated by the shared environment so that the agents' objectives and the global system specifications are sattisfied. In this talk we will view agents interactions as a rule governed activity which is modeled as a game. Games consist of a set of moves, their effects and the initial and the final state of a game. We will show how a games metaphor provides an intuitive model for representing agent interactions in distributed platforms. On one hand, the cooperations in the agent environment result from applying set of coordination games. On the other hand, given the specification of a game, agents are enabled to reason about the rules and build strategies to bring about success in their individual goals. We will show how we applied the coordination games techniques to different large-scale problems and discuss open issues and future developments.
evento contenitore
