Research Project

Machine Learning and Intelligent Agents Logo and Link

Machine Learning and
Intelligent Agents

Trading Agents


Maria Fasli

Project Description:

Electronic commerce (e-commerce) is the use of the Internet to conduct commerce. Agent-based systems have proven useful in helping individuals shop for consumer items, like air-line tickets, CDs and books as well as businesses to conduct commerce over the Internet both with consumers and other businesses. Agents are particularly well-suited for e-commerce applications since they are semi-autonomous pieces of software that act on behalf of their user (individual or organisation) and according to their preferences. This project explores the use of agent technology for e-commerce.

As part of this project we have been developing agents for the International Trading Agent Competition (TAC) which provides a benchmark for the performance of complicated trading agents in a challenging market game. In the 2001 competition we achieved the 7th place with the agent CaiserSose competing against 28 teams from research and academic institutions as well as companies from all around the world. In the 2002 competition we achieved the 3rd place with the agent Thalis competing against 26 other teams. In 2003 Thalis obtained the 4th position. We have also been developing strategies for the new Supply Chain Management game and in the ICEC 2003 competition we obtained the 7th place with our agent Socrates.

More about the TAC competition can be found at:

[top of page]


  Tuesday, 26 October 2004