The topic of the first paper in this issue is CMieux, CMU's Trading Agent Competition (TAC) Supply Chain Management (SCM) agent. Intended for dynamic (rather than static) supply chains, CMieux is designed to rapidly adapt to changing market conditions. The paper shows how CMieux outperformed other agents on the procurement side, and matched their bidding performance.
Electronic commerce transactions are often anonymous. As a consequence, participants cannot be assumed trustworthy. Instead, mechanisms must be designed that implement trust. The second paper in this issue describes a reputation mechanism for layered peer-to-peer communities that establishes trust. A simulation was created and experiments were run to validate the design.
In the third paper, the winner determination problem in auctions is framed as one of exploration, rather than optimization, and a "lightweight" algorithm is presented that outputs the k-best solutions. Applied to (reverse) procurement auctions, the algorithm scales to practical problem sizes.
This issue of SIGecom Exchanges includes a second paper on the TAC, about TAC Travel, rather than TAC SCM. This paper focuses on the optimization piece of the design of TAC Travel agents, addressing the decision-theoretic bidding problem analytically, and the game-theoretic bidding problem experimentally.
Amy Greenwald
Editor-in-Chief