SIMPLE - A Multi-Agent System for Simultaneous and Related Auctions

This work describes a multi-agent architecture and strategy for trade in simultaneous and related auctions. The proposed SIMPLE Agency combines an integer programming model, machine learning techniques and knowledge engineering strategies to build its decision making process. Some well known problems in trading were identified in order to design the architecture. Each SIMPLE agent is concerned with one of the trading subproblems. This makes it possible to apply different computational techniques to solve each subproblem separately and then combine their solutions. The Trading Agent Competition (TAC) is used to illustrate and test our approach. SIMPLE shows high performance on competitive scenarios using the TAC server environment.

• Type: Conference
• Year: 2003
• Where: IAT
• Authors: Ruy Luiz Milidiú, Taciana Melcop, Frederico dos S. Liporace, Carlos J. Pereira de Lu
• Keywords: IAT 2003, Integer Programming Model, Intelligent Agent, Machine Learning Techniques, Trading Agent Competition