Agent-based bidding in electronic auctions

University Of Tulsa, Tulsa OK

Investigators

Abstract

This project will investigate bidding strategies using agents which can buy goods and services from multiple auctions while satisfying user preferences and budgetary constraints. Of particular interest is the study and development of bundle bidding strategies, which will allow agents to put together preferred bundles of goods for users from online auctions selling only individual items. Automated bundle bidding procedures with well-understood performance expectations will allow users to form bundles with significantly less effort and much more success. The bundle bidding problem presents challenging and interesting research problems that involve asynchronous events, uncertainty about future auctions, and the widely varying valuations and bidding strategies of users and their proxies. Several variations of the bundle-bidding problem will be identified and studied. The project will also investigate the usage of auctions in conjunction with other forms of information processing within a broader context of coordination between self-interested autonomous agents representing individuals, organizations and supply chains.