2000
@techreport{MMS2000,
vgclass = {report},
vgproject = {viper,cbir},
author = {Henning M\"{u}ller and Wolfgang M\"{u}ller and David
McG. Squire and St\'{e}phane Marchand-Maillet and Thierry Pun},
title = {Strategies for positive and negative relevance feedback in
image retrieval},
number = {00.01},
institution = {Computer Vision Group, Computing Centre, University of
Geneva},
address = {rue G\'{e}n\'{e}ral Dufour, 24, CH-1211 Gen\`{e}ve, Switzerland},
month = {January},
year = {2000},
url = {/publications/postscript/2000/VGTR00.01_MuellerHMuellerWSquireMarchandPun.pdf},
url1 = {/publications/postscript/2000/VGTR00.01_MuellerHMuellerWSquireMarchandPun.ps.gz},
abstract = {Relevance feedback has been shown to be a very
effective tool for enhancing retrieval results in text retrieval. In
content-based image retrieval it is more and more frequently used and
very good results have been obtained. However, too much negative
feedback may destroy a query as good features get negative weightings.
This paper compares a variety of strategies for positive and negative
feedback. The performance evaluation of feedback algorithms is a hard
problem. To solve this, we obtain judgments from several users and
employ an automated feedback scheme. We can then evaluate different
techniques using the same judgments. Using automated feedback, the
ability of a system to adapt to the user's needs can be measured very
effectively. Our study highlights the utility of negative feedback,
especially over several feedback steps.},
}