Abstract We propose a new scheme of `automatic pricing' for digital contents, and describe an implemented system as well as concrete pricing algorithms for it. Automatic pricing refers to a methodology of automatically setting sales prices to optimal prices, based on past prices and sales. In particular, we consider the case in which automatic pricing is done in order to maximize the profit of an on-line marketing site. We describe a demo site for on-line marketing with automatic pricing, which we call `digiprice'. We will also describe the concrete pricing algorithms we employ in digiprice, and report on preliminary performance evaluation experiments we conducted using simulated data. The results of experimentation verify that our methods are practical in terms of both the speed of convergence to the optimal price and computational efficiency.
[1]
Philip M. Long,et al.
Apple tasting and nearly one-sided learning
,
1992,
Proceedings., 33rd Annual Symposium on Foundations of Computer Science.
[2]
F. Downton.
Stochastic Approximation
,
1969,
Nature.
[3]
Philip M. Long,et al.
Associative Reinforcement Learning using Linear Probabilistic Concepts
,
1999,
ICML.
[4]
Philip M. Long.
On-line evaluation and prediction using linear functions
,
1997,
COLT '97.
[5]
Naoki Abe,et al.
The “lob-pass” problem and an on-line learning model of rational choice
,
1993,
COLT '93.
[6]
Leslie Pack Kaelbling,et al.
Associative Reinforcement Learning: Functions in k-DNF
,
1994,
Machine Learning.