Evaluating user preferences of web search results is crucial for search engine development, deployment, and maintenance. We present a real-world study of modeling the behavior of web search users to predict web search result preferences. Accurate modeling and interpretation of user behavior has important applications to ranking, click spam detection, web search personalization, and other tasks. Our key insight to improving robustness of interpreting implicit feedback is to model query-dependent deviations from the expected "noisy" user behavior. We show that our model of clickthrough interpretation improves prediction accuracy over state-of-the-art clickthrough methods. We generalize our approach to model user behavior beyond clickthrough, which results in higher preference prediction accuracy than models based on clickthrough information alone. We report results of a large-scale experimental evaluation that show substantial improvements over published implicit feedback interpretation methods.
Description
Learning user interaction models for predicting web search result preferences
%0 Conference Paper
%1 Agichtein06
%A Agichtein, Eugene
%A Brill, Eric
%A Dumais, Susan
%A Ragno, Robert
%B SIGIR '06: Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval
%C New York, NY, USA
%D 2006
%I ACM
%K UserModel WebSearch prediction
%P 3--10
%R http://doi.acm.org/10.1145/1148170.1148175
%T Learning user interaction models for predicting web search result preferences
%U http://portal.acm.org/citation.cfm?id=1148170.1148175
%X Evaluating user preferences of web search results is crucial for search engine development, deployment, and maintenance. We present a real-world study of modeling the behavior of web search users to predict web search result preferences. Accurate modeling and interpretation of user behavior has important applications to ranking, click spam detection, web search personalization, and other tasks. Our key insight to improving robustness of interpreting implicit feedback is to model query-dependent deviations from the expected "noisy" user behavior. We show that our model of clickthrough interpretation improves prediction accuracy over state-of-the-art clickthrough methods. We generalize our approach to model user behavior beyond clickthrough, which results in higher preference prediction accuracy than models based on clickthrough information alone. We report results of a large-scale experimental evaluation that show substantial improvements over published implicit feedback interpretation methods.
%@ 1-59593-369-7
@inproceedings{Agichtein06,
abstract = {Evaluating user preferences of web search results is crucial for search engine development, deployment, and maintenance. We present a real-world study of modeling the behavior of web search users to predict web search result preferences. Accurate modeling and interpretation of user behavior has important applications to ranking, click spam detection, web search personalization, and other tasks. Our key insight to improving robustness of interpreting implicit feedback is to model query-dependent deviations from the expected "noisy" user behavior. We show that our model of clickthrough interpretation improves prediction accuracy over state-of-the-art clickthrough methods. We generalize our approach to model user behavior beyond clickthrough, which results in higher preference prediction accuracy than models based on clickthrough information alone. We report results of a large-scale experimental evaluation that show substantial improvements over published implicit feedback interpretation methods.},
added-at = {2008-11-21T11:40:20.000+0100},
address = {New York, NY, USA},
author = {Agichtein, Eugene and Brill, Eric and Dumais, Susan and Ragno, Robert},
biburl = {https://www.bibsonomy.org/bibtex/23954a57d76e949b91be0967a26f3842a/mkroell},
booktitle = {SIGIR '06: Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval},
description = {Learning user interaction models for predicting web search result preferences},
doi = {http://doi.acm.org/10.1145/1148170.1148175},
interhash = {f518cdbd590b7e54fb1c2f8eedc68359},
intrahash = {3954a57d76e949b91be0967a26f3842a},
isbn = {1-59593-369-7},
keywords = {UserModel WebSearch prediction},
location = {Seattle, Washington, USA},
pages = {3--10},
publisher = {ACM},
timestamp = {2009-08-06T09:25:46.000+0200},
title = {Learning user interaction models for predicting web search result preferences},
url = {http://portal.acm.org/citation.cfm?id=1148170.1148175},
year = 2006
}