An MDE Approach for Modelling and Reasoning About Multi-agent Systems

摘要

Epistemic logic plays an important role in artificial intelligence for reasoning about multi-agent systems. Current approaches for modelling multi-agent systems with epistemic logic use Kripke semantics where the knowledge base of an agent is represented as atomic propositions, but intelligent agents need to be equipped with formulas to derive implicit information. In this paper, we propose a metamodelling approach where agents' state of affairs are separated in different scopes; and the knowledge base of an agent is represented by a propositional logic language restricted to Horn clauses. We propose to use a model driven approach for the diagrammatic representation of multi-agent systems knowledge (and nested knowledge). We use a message passing for updating the state of affairs of agents and use belief revision to update the knowledge base of agents.