Applying Multi-Agent Algorithm to a Class Scheduling System

Abstract

We propose a multi-agent algorithm to solve a scheduling problem. The proposed application takes into account various restrictions relative to Japanese university class scheduling in particular. A multi-agent system consisting of agents representing the requirements and restrictions for professors and departments are introduced. Scheduling is solved using negotiation between agents together with the hill-climbing method. Through a software simulation, we were able to show that our proposed method successfully applies the self-organizing nature of agents to solve the scheduling problem.