Specification by refinement and agreement: Designing agent interaction using landmarks and contracts

Abstract

In this paper, we argue that multi-agent systems developed to model and support organizations must on the one hand be able to describe and realize the goals and structure of that particular organization and on the other hand allow for autonomous behavior of participating actors. The agent autonomy should include collaboration autonomy, which means that agents can decide among each other how they want to cooperate. We present a model for agent societies and two techniques that can be used to achieve the described objectives: landmarks and contracts. The effects of these techniques on different agent society types are explored as well.