Executable specication of open multi-agent systems

Abstract

Multi-agent systems where the agents are developed by parties with competing interests, and where there is no access to an agent's internal state, are often classi ed as `open'. The members of such systems may inadvertently fail to, or even deliberately choose not to, conform to the system speci cation. Consequently, it is necessary to specify the normative relations that may exist between the members, such as permission, obligation, and institutional power. We present a framework being developed for executable speci cation of open multi-agent systems. We adopt a bird's eye view of these systems, as opposed to an agent's perspective whereby it reasons about how it should act. This paper is devoted to the presentation of various examples from the NetBill protocol formalised in terms of institutional power, permission and obligation. We express the system speci cation in the Event Calculus and execute the speci cation by means of a logic programming implementation. We also give several example formalisations of sanctions for dealing with violations of permissions and obligations. We distinguish between an open multi-agent system and the procedure by which an agent enters and leaves the system. We present examples from the speci cation of a role-management protocol for NetBill, and demonstrate the interplay between such a protocol and the corresponding multi-agent system.