A Modular Architecture for Integrating Normative Advisors in MAS

This paper introduces a modular architecture for integrating norms in autonomous agents and multi-agent systems. As the interactions between norms and agents can be complex, this architecture utilizes multiple programmable components to model concepts such as adoption of personal and/or collective norms (possibly conflicting), interpretation and qualification as mappings between social and normative contexts, intentionally (non-)compliant behaviors, and resolution of conflicts between norms and desires (or other norms). The architecture revolves around normative advisors, that act as the bridge between intentional agents and the institutional reality. As a technical contribution, a running implementation of the architecture is presented based on the ASC2 (AgentScript) BDI framework and eFLINT normative reasoner.