473 A Classification of Auction Mechanism: Potentiality for Multi-Agent System (MAS) Based Modeling Anurag Jain University of Texas at Arlington Arlington, Texas 817-272.3502 Fax: 817-272.5801 Email: ajain@uta.edu Riyaz Sikora University of Texas at Arlington Arlington, Texas 817-272.3502, Fax: 817-272.5801 Email: rsikora@uta.edu ABSTRACT What are auctions and how are they applicable in Multi-agent systems (MAS). This study reviews the various auction mechanisms such as forward auctions, reverse auctions, and double auctions, along with the roots of the newer auctions that have been evolved over time. Subsequently, a classification of various auction mechanism are presented that will enable us to get a perspective on how auctions have evolved and thereby a better understanding on how to incorporate these mechanism into a MAS. Later, important auction properties are compared with the auction mechanism to get a better grasp of what a particular action might possess as strength and its related weakness. Auction mechanism have been applied in agent based modeling. This study illustrates the comprehensive set of current and emerging auctions mechanism that have the potentially to be tested through agent based modeling as well as serve as a basis for further development of Multi-agent systems. INTRODUCTION Auctions have been an important area of study for while. Some of the earlier work in traditional auctions can be found in Economic research [McAfee and McMillian, 1987, Milgrom, 1989]. With the importance to the study of auctions, there have been attempts to propose a theory of auctions too [Milgrom, 1982]. The differentiation between negotiations and auctions is that auctions have protocols that are enforced. This ensures transparency of the transaction and thereby lets the members of the auction feel that the process is not unfair [Pinker et al., 2003]. In the setting of intelligent agent study and Multi-agent systems (MAS), auctions provide coordination mechanisms for improved negotiation that can be emulated by agents. Therefore, this makes this an compelling area of research to incorporate agent based modeling.