Agentplanner - Agent-Based Timetabling System - Preliminary Design and Evaluation

The aim of this note is to present the initial design and preliminary evaluation of the AgentPlanner, an agent-based timetabling system. The primary advantage of the agent approach is the relative ease of schedule modification. This is particularly important in the proposed application area: scheduling university courses. Experimental results, comparing the performance of the AgentPlanner with a state-of-the-art genetic algorithms based software, obtained for a single department and a single building, are presented and analyzed.