Modelling Agents and Communication using CommonKADS

Abstract

A number of systematic methods have been developed to formalise and direct the knowledge engineering process. One such method is the CommonKADS methodology, which consists of six models capturing knowledge at different levels of abstraction. This paper examines two of these six models: the Agent model and the Communication model. The Agent and Communication models are responsible for modelling “co-operation” within CommonKADS. The Agent model majors on who has the capability to carry out each task, while the Communication model highlights where information is needed within the process (i.e. which agent holds the information that forms the input and output of key tasks).

This paper will propose that both the Agent and Communication models can usefully be developed at two levels of detail: the level of tasks within a business process (between-tasks), and the level of subtasks (within-task). By developing both models at both levels, a comprehensive set of information is obtained. Extensions and adaptations are proposed to both models, first to ensure full and accurate knowledge representation, and then to discuss how a “shorthand” version of the models might be developed.