Programming MAS with Artifacts

This paper introduces the notion of artifact as a first-class abstraction in MASs (multi-agent systems) and focuses on its impact on MAS programming. Artifacts are runtime devices providing some kind of function or service which agents can fruitfully use—both individually and collectively—to achieve their individual as well as social objectives. Artifacts can be conceived (and programmed) as basic building blocks to model and build agent (working) environments. Besides introducing a conceptual and modelling framework, the paper discusses the impact of this new notion on MAS programming, focussing in particular on MAS composed by cognitive agents. To make the discussion more concrete, we provide an example scenario featuring 3APL agents whose coordination activity is supported by TuCSoN tuple centres—an existing coordination model providing some of the basic properties of artifacts for MASs.