A decision-theoretic algorithm for bundle purchasing in multiple open ascending-price auctions

From National Research Council Canada

DOIResolve DOI: http://doi.org/10.1007/978-3-540-24840-8_32
AuthorSearch for: 1 ; Search for: 2 
Name affiliation
  1. National Research Council Canada. Information and Communication Technologies
  2. National Research Council Canada. NRC Institute for Information Technology
TypeBook Chapter
Proceedings titleAdvances in Artificial Intelligence
Series titleLecture Notes in Computer Science; Volume 3060
Conference17th Conference of the Canadian Society for Computational Studies of Intelligence, May 17-19, 2004, London, Ontario, Canada
ISSN0302-9743
1611-3349
ISBN978-3-540-22004-6
978-3-540-24840-8
Pages429–433
Abstract
This paper presents an algorithm for decision-making where the buyer needs to procure one of possibly many bundles of complementary products, and items are sold individually in multiple open ascending-price (English) auctions. Auctions have fixed start and end times, and some may run concurrently. Expected utility of a bidding choice is computed by considering expected utility of choices at future decisions. The problem is modeled as a Markov decision process, and dynamic programming is used to determine the value of bidding/not bidding in each state. Three techniques for reducing the state space are given. Results show that a bidding agent using this algorithm achieves significantly higher utility on average when compared with one that does not consider the value of future choices.
Publication date
PublisherSpringer Berlin Heidelberg
LanguageEnglish
Peer reviewedYes
NPARC number23000092
Export citationExport as RIS
Report a correctionReport a correction
Record identifier0efb30e1-7e34-4682-99a5-50e5b0e4633a
Record created2016-06-02
Record modified2019-01-10
Date modified:
Report a problem on this page