On the automatic verification of interaction protocols using g−SCIFF

Interaction protocol verification has been in recent years intensively investigated within and outside multi-agent research. Many techniques and models have been proposed based on a number of assumptions and choices, among which are the kind of knowledge available and the kind of properties that are subject of verification. We focus on interaction protocol verification in open multi-agent systems, where the internal architecture of the individual agents is not known a-priori. In such a setting, while it is not possible to predict the behaviour of individual agents, it is still possible to approach the interaction verification problem either by assuming that agents will indeed comply to the protocols, thus focussing on the study of properties of the interaction protocols, or by assuming that the agent behaviour, if not predictable, is at least observable, and then focussing on the run-time verification of compliant behaviour of interacting agents by reasoning on the messages they exchange with each other. In this article, we build on previous work on logic-based specification and run-time verification of protocol compliance, and we extend the SOCS social framework towards verification of protocol properties. In doing so, we propose a unified formal and computational framework based on abductive logic programming which can be used to (i) design and specify interaction protocols that provably exhibit desirable properties and (ii) verify, at run-time, that agent interaction actually complies to the specified interaction protocols.