Designing a Multiagent System for Course-Offering Determination

Abstract

This paper describes the design of a multiagent system that facilitates course-offering decision making for a program in an institution. We first model course offering determination for upcoming semester as a multi-winner election with exogenous constraints which is a problem for computational social choice in multiagent systems, which has rarely been considered. Then, the paper describes the architecture and models of the multiagent system for course offering determination with Gaia role model methodology, TROPOS strategic actor diagram, Agent Unified Modeling Language (AUML) sequence diagram for a multi-agent negotiation interaction protocol, and Pseudo-code algorithms for generating fractional votes and course election protocol. A novel course selection preference model for students has been proposed and described formally. The effectiveness of the approach and the implemented system has been showed with the initial experimental results.