Decision-Theoretic Learning of Agent Models in an Information Economy

Christopher H. Brooks and Edmund H. Durfee

We demonstrate how a producer of information goods can use a successively complex series of models to learn the preferences of consumers efficiently. We provide metrics for estimating the precision, accuracy, and learning complexity of different models, thereby providing a producer with the metrics needed to apply decision theory in selecting a sequence of models. We present experimental results demonstrating the effectiveness of this approach, and discuss current research on extending this idea to learning preferences over categories or strategies of other agents.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.