Improving E-Trade Auction Volume by Consortium - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 92

The environment of agent-based auction trading involves online auctions with Web-based buyers
and Web-based sellers, typically multiple buyers and multiple suppliers. Therefore how to improve the
performance of agent-based auction, specifically match buyers and suppliers, is a challenge for both
academia and practitioners. In practice, there are two major parameters of a proposed trade, the price
and the volume. This is equally true for a seller's offer or for a buyer's bid. Typically, volumes larger
than certain lower levels go with lower prices. Thus, a match between a seller's offer and a buyer's bid
that would lead to a transaction is a combination of matches with volume and price. In this research,
we only consider the match of volume, since MRO products usually have a prevailing marketing price,
and thus price is automatically matched. Volume is subject to individual agent's bids and offers in a
buyer and seller context, and when a direct match is not possible, a match may still be obtained by
cooperation/consortium between buyers and sellers, respectively. The focus of this chapter is on the
effectiveness of cooperation in making the match and completing the trade.
In this chapter, we propose a two-tier e-procurement auction agent structure made up of multiple
suppliers and multiple buyers. Trading begins with a buyer proposing a trading amount. A seller may
match the trading amount, or may propose a different trading amount. The buyer then seeks to match
the trading amount with cooperation from other buyers. Alternately, the seller seeks to suit the buyer, or
hold the order for a future offer that matches. The purpose of this approach is to provide better matches
with offers, while reducing wait periods by means of cooperative trading. Thus, the efficiency of trading
is increased.
When agents work together within a community, collaborating to achieve individual goals, it be-
comes a multi-agent system (MAS), where the interactions between the agents become as important
as the decision-oriented actions of the individual agents. At this time, some research has been done on
agent cooperation in multi-agent systems. For example, Binbasioglu (1999) proposed an approach to
identify problem components, which supports the progress of understanding and structuring for multi-
agent cooperative decision making environment. Fox et al. (2000) presented a solution to construct
agent-oriented software architecture to manage supply chain at tactical and operational levels. In their
framework, they used multiple agents, such as order acquisition agents, logistics agents, transportation
agents, scheduling agents, resources agents, etc. One important capability of their agents, related to the
present work, is the coordination. The authors developed a generic application-independent language
to implement the multi-agent coordination issue. Kosakaya et al. (2001) developed a new cooperation
and negotiation algorithm to improve cooperation in a system using multi-agent system. Zhao et al.
(2001) developed the agent-based CLOVER platform that can improve system interoperability among
agents, and furthermore support dynamic and flexible cooperation. Aknine et al. (2004) proposed two
methods of agents' coalition formation for both cooperative and non-cooperative multi-agent systems,
and cooperative agents can exchange information/preferences among them. Based on the virtual enter-
prise (VE) paradigm and the concept of multi-agent, Roy et al. (2004) proposed a new way to manage
supply chains. They defined tiered supply chain architecture, where a virtual enterprise node (VEN)
only interacts with an adjacent VEN. The objective is to coordinate the decentralized VEN decisions
in real time, and each VEN needs to make a tradeoff between local benefits and global benefits. Anus-
sornnitisarn et al. (2005) developed a model of distributed network for distributed resource allocation,
and they investigated if multi-agent system, as a whole, can achieve efficient resource allocation in a
collaborative environment. Hill et al. (2005) designed a cooperative multi-agent approach for decentral-
ized decision-making environment in free flight, which provides effective results for different scenarios.
Zho et al. (2006) applied intelligent multi-agent technology in manufacturing systems, where agents