A high level visual notation for understanding and designing collaborative, adaptive behaviour in multiagent systems

The authors explain how to use a high level, visual notation called use case maps (UCMs) to bring together the “what” and “how” of multi-agent systems for understanding and design. “What” refers to descriptions of what agents do, descriptions that are declarative from a detailed software design perspective (e.g., BDI models). “How” refers to descriptions of how the software does it expressed with software design notations. Two important properties of agent systems that make them difficult to understand and design are multi-agent collaborative behaviour and adaption by system self modification. BDI-style descriptions of what must be done to achieve these properties do not give a direct view of the properties, but leave them to emerge. Conventional software design descriptions swamp us with unnecessary and undesirable derail relative to both system properties and BDI-style models. UCMs were invented to raise the level of abstraction of software design in precisely the way required to overcome these problems. UCMs are particularly suitable for multi-agent systems because they bring together the “what” and “how” of collaborative behaviour and system self modification in a coherent way, in a single high level visual notation. A companion HICSS´98 paper titled “Applying Use Case Maps to Multi-agent Systems: A Feature Interaction Example” illustrates the approach