Towards adaptive normative systems for communities of agents

Abstract

We are concerned with normative systems for communities of human and artificial agents that cooperate for improving their overall goals. Unlike open artificial multi-agent systems, we need to be more flexible in judging the behavior of the agents, and still thrive for efficiency. To attain the overall system goals norms are used for guiding the behavior of the agents, whether human or artificial. To provide flexibility in adaptation the power of ontology expressed in description logic is used. The problem consists then in conceiving some mechanism that can support flexibility for the system, without imposing a too heavy burden for such applications. Some cases are presented to show how our proposal satisfies the needs for adapting and evolving normative systems for communities of practice. The interplay of the norms and the concepts in the ontology provides a period of grace for minor offenders, wether artificial or human.