Content:
In this paper the issue of bidding strategy learning in electronic markets is addressed. The primary aim is to identify machine learning techniques which are best suited to learn bidding behaviour in electronic markets. The developed methodologies are applied within a structured market engineering process to improve the quality of market designs. Market simulations are carried out based on a discriminatory price double auction market design. The simulations of the market behavior are modelled in a multi-agent system. The market participants which are (i) market maker agents (ii) supplier agents (iii) consumer agents act as autonomous agents on a simulated market. Supplier agents are static, i.e. not equipped with learning techniques. Consumer agents are modelled using different machine learning methods for price determination. We develop the market simulation library MELBOURNE (Market Engineering Library for Bidding Objectives Using Realtime Negotiation Environments). In the simulation runs using the library, we show that learning algorithms always outperform heuristic methods in price discovery. In: Poster Proceedings of the Workshop on Agent Mediated Electronic Commerce V (AMEC-V), held at the Second International Joint Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), Melbourne, 2003

Literature: