Learning to Advertise: How Many Ads Are Enough?

Sponsored advertisement(ad) has already become the major source of revenue for most popular search engines. One fundamental challenge facing all search engines is how to achieve a balance between the number of displayed ads and the potential annoyance to the users. Displaying more ads would improve the chance for the user clicking an ad. However, when the ads are not really relevant to the users' interests, displaying more may annoy them and even "train" them to ignore ads. In this paper, we study an interesting problem that how many ads should be displayed for a given query. We use statistics on real ads click-through data to show the existence of the problem and the possibility to predict the ideal number. There are two main observations: 1) when the click entropy of a query exceeds a threshold, the CTR of that query will be very near zero; 2) the threshold of click entropy can be automatically determined when the number of removed ads is given. Further, we propose a learning approach to rank the ads and to predict the number of displayed ads for a given query. The experimental results on a commercial search engine dataset validate the effectiveness of the proposed approach.

[1]  Nick Craswell,et al.  An experimental comparison of click position-bias models , 2008, WSDM '08.

[2]  Andreas Krause,et al.  Online Learning of Assignments , 2009, NIPS.

[3]  James G. Shanahan,et al.  Boosting support vector machines for text classification through parameter-free threshold relaxation , 2003, CIKM '03.

[4]  Ji-Rong Wen,et al.  A large-scale evaluation and analysis of personalized search strategies , 2007, WWW '07.

[5]  Jimeng Sun,et al.  Social influence analysis in large-scale networks , 2009, KDD.

[6]  Tie-Yan Liu,et al.  Actively predicting diverse search intent from user browsing behaviors , 2010, WWW '10.

[7]  Manish S. Gupta.  Predicting click through rate for job listings , 2009, WWW '09.

[8]  Andrei Z. Broder,et al.  To swing or not to swing: learning when (not) to advertise , 2008, CIKM '08.

[9]  Wei Fan,et al.  Heterogeneous cross domain ranking in latent space , 2009, CIKM.

[10]  Stephen E. Robertson,et al.  Where to stop reading a ranked list?: threshold optimization using truncated score distributions , 2009, SIGIR.

[11]  Filip Radlinski,et al.  Optimizing relevance and revenue in ad search: a query substitution approach , 2008, SIGIR '08.

[12]  Dawid Weiss,et al.  Predicting Ads' Click-Through Rate with Decision Rules , 2008 .

[13]  Benjamin Piwowarski,et al.  A user browsing model to predict search engine click data from past observations. , 2008, SIGIR '08.

[14]  Deepak Agarwal,et al.  Spatio-temporal models for estimating click-through rate , 2009, WWW '09.

[15]  Ben Carterette,et al.  Evaluating Search Engines by Modeling the Relationship Between Relevance and Clicks , 2007, NIPS.

[16]  Matthew Richardson,et al.  Predicting clicks: estimating the click-through rate for new ads , 2007, WWW '07.

[17]  Qiang Wu,et al.  Click-through prediction for news queries , 2009, SIGIR.

[18]  Olivier Chapelle,et al.  A dynamic bayesian network click model for web search ranking , 2009, WWW '09.

[19]  Zheng Chen,et al.  A novel click model and its applications to online advertising , 2010, WSDM '10.

[20]  John F. Canny,et al.  Large-scale behavioral targeting , 2009, KDD.

[21]  Chao Liu,et al.  Click chain model in web search , 2009, WWW '09.

[22]  Jian Hu,et al.  Optimizing search engine revenue in sponsored search , 2009, SIGIR.

[23]  Vassilis Plachouras,et al.  Online learning from click data for sponsored search , 2008, WWW.