Decision Making with a KGP Agent System

A human user is centrally involved in a Decision Support System (DSS) task, whereas for an autonomous agent system, a user’s role is a more abstract notion of oversight and approval. In this paper, we examine the potential integration of a multi-agent architecture into a DSS. The motivation in this work is to utilise autonomous agent problemsolvers within a suitable multi-agent framework in such a way as to realise and generalise the capabilities of a DSS, while maintaining the DSS characteristic of significant user involvement. The approach taken is to employ agents both as domain problem solvers as well as to manage the interaction with the user as part of the semi-autonomous design. We consider the construction of an agent-based DSS based upon the KGP model of agency and detail how the design of a DSS may benefit from agent technology in general and KGP agents in particular (and vice versa). A framework for agent-enhanced DSS is presented which is more general than those presented previously. We show how the architectural framework used for KGP agents can be specialised to realise enhanced DSS capabilities. RÉSUMÉ. L’ensemble des.