Improving E-Trade Auction Volume by Consortium - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 99

feasibilities of the transactions, and cost considerations associated with remainder quantities which have
to be sold at lower prices with higher transaction costs leading to higher prices for a buyer agent.

Procedures of agent t rading

As described above, the Dynamic Programming algorithm is an elegant and correct representation of the
trading conditions, but is difficult to solve and apply in practice. We describe below a heuristic approach
to solving the decision problem in agent trading. In this process, we describe the architecture of the
agent community involved in trading and the interactions between the agents involved in the transac-
tions. Presented in Figure 4 is a Petri Net representation of the agent-trading model. The representations
correspond to the conventions of a Petri Net representation, with a circle designated “ P n ” representing
a decision state in the trading actions performed by an agent, and a box designated “ t n ” representing an
action, such as an offer, an acceptance or rejection of an offer, or a request for help from other members
of the community. The arrows represent transitions between decision states, or actions, induced by the
occurrences of events. The complete diagram is a representation of all possible states and transitions
encountered in the environment of auction trading by agents.
The agent structure described in Figure 1 leads to the decision states and events of auction trans-
actions in Figure 4. A more specific set of definitions of the decision states, and the events that lead to
transitions to other decision states is given below:

P0 : Start
P1 : Buyer's trading decision state
P2 : Buyer's no trading state
P3 : Buyer's trading state
P4 : Sellers' consortium state
P5 : Seller's trading decision state
P6 : Seller's no trading state
P7 : Seller's trading state
P8 : Buyers' consortium state

t 1 : Buyer offer & seller ask
t 2 : Buyer rejects the seller's offer
t 3 : Buyer accepts the seller's offer
t 4 : Buyers form a buyer's consortium
t 5 : Buyers offer jointly
t 6 : Seller ask & buyer offer
t 7 : Seller rejects the buyer's offer
t 8 : Seller accepts the buyer's offer
t 9 : Seller forms a seller's consortium
t 10 : Sellers offer jointly

With these definitions, and with the understanding that arrows represent times to an event and a
state transition, the diagram in Figure 4 explains itself. However, the dynamic programming model in