Inferring What a User is Not Interested In
- Robert Holte, Department of Computing Science, University of Alberta
- John Ng Yuen Yan, BNR Ltd., Ottawa
This paper describes a system to improvethe speed and success rate with whichusers browse softwarelibraries. The system is a learning apprentice: it monitorsthe user's normal browsing actions and from these infersthe goal of the user'ssearch.Itthen searches the library being browsed, uses the inferred goal to evaluate items and presents to the user those that aremost relevant. The main contribution of this paper is the development of rules for negative inference (i.e.inferring features that the user is not interested in). These produce a dramatic improvement in the system'sperformance. The newsystem is morethan twice as effective at identifying the user'ssearch goal than the original, and it ranks the targetmuchmoreaccurately at all stagesofsearch.
Citation
R. Holte, J. Yan. "Inferring What a User is Not Interested In". Canadian Conference on Artificial Intelligence (CAI), pp 159-171, May 1996.| Keywords: | inferring, improvement, performance |
| Category: | In Conference |
BibTeX
@incollection{Holte+Yan:CAI96,
author = {Robert Holte and John Ng Yuen Yan},
title = {Inferring What a User is Not Interested In},
Pages = {159-171},
booktitle = {Canadian Conference on Artificial Intelligence (CAI)},
year = 1996,
}Last Updated: June 18, 2007Submitted by Staurt H. Johnson
