A Belief-Based Multi-Agent Markov Decision Process for Staff Management

This paper presents a system designed for task allocation, staff management and decision support in a large enterprise, in which permanent staff and contractors work alongside under the overall management of a manager to handle tasks initiated by end-users. The process of allocating a new task to a worker is modeled under different situations, taking into account user requirements as well as the different goals of management, permanent staff and contractors. Their actions and strategies are formalized as autonomous decision-support subsystems inside a multi-agent system, based on Contract Net Protocol, belief theory, multiobjective optimization theory and Markov Decision Process.