Evaluating Retrieval Performance Using Clickthrough Data
T. Joachims. page 79--96. Physica/Springer Verlag, (2003)
Abstract
This paper proposes a new method for evaluating the quality of retrievalfunctions. Unlike traditional methods that require relevance judgmentsby experts or explicit user feedback, it is based entirely on clickthroughdata. This is a key advantage, since clickthrough data can becollected at very low cost and without overhead for the user. Takingan approach from experiment design, the paper proposes an experimentsetup that generates unbiased feedback about the relative quality of two...
%0 Book Section
%1 citeulike:460217
%A Joachims, T.
%B Text Mining
%D 2003
%E Franke, J.
%E Nakhaeizadeh, G.
%E Renz, I.
%I Physica/Springer Verlag
%K clickstream evaluation implicit-feedback
%P 79--96
%T Evaluating Retrieval Performance Using Clickthrough Data
%U http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.4.1591
%X This paper proposes a new method for evaluating the quality of retrievalfunctions. Unlike traditional methods that require relevance judgmentsby experts or explicit user feedback, it is based entirely on clickthroughdata. This is a key advantage, since clickthrough data can becollected at very low cost and without overhead for the user. Takingan approach from experiment design, the paper proposes an experimentsetup that generates unbiased feedback about the relative quality of two...
@inbook{citeulike:460217,
abstract = {{This paper proposes a new method for evaluating the quality of retrievalfunctions. Unlike traditional methods that require relevance judgmentsby experts or explicit user feedback, it is based entirely on clickthroughdata. This is a key advantage, since clickthrough data can becollected at very low cost and without overhead for the user. Takingan approach from experiment design, the paper proposes an experimentsetup that generates unbiased feedback about the relative quality of two...}},
added-at = {2017-11-15T17:02:25.000+0100},
author = {Joachims, T.},
biburl = {https://www.bibsonomy.org/bibtex/20d249fba117e983e78ba72e878988e25/brusilovsky},
booktitle = {Text Mining},
citeulike-article-id = {460217},
citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.4.1591},
editor = {Franke, J. and Nakhaeizadeh, G. and Renz, I.},
interhash = {a0aaf0651078d3fadc8b83a9d0310212},
intrahash = {0d249fba117e983e78ba72e878988e25},
keywords = {clickstream evaluation implicit-feedback},
pages = {79--96},
posted-at = {2009-07-16 17:14:30},
priority = {2},
publisher = {Physica/Springer Verlag},
timestamp = {2020-06-25T15:52:02.000+0200},
title = {{Evaluating Retrieval Performance Using Clickthrough Data}},
url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.4.1591},
year = 2003
}