A survey on click modeling in web search


 Jody Scott
 2 years ago
 Views:
Transcription
1 A survey on click modeling in web search Lianghao Li Hong Kong University of Science and Technology
2 Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
3 Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
4
5 Search engine marketing
6 Generalized secondprice auction
7 Search advertising demo
8 Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
9 Why do we need click prediction? Revenue is highly influenced by click probability prediction. Search engines rank ads with expected revenue E[revenue] = P ad (click) GSP(ad)
10 How to predict click behavior? Clickthrough logs help! Figure: Ranking presented for the query support vector machine
11 How to predict click behavior? To predict clicks by counting! P ad (click) = # of clicks # of impressions However, that is far from satisfaction clicks are biased due to the user browsing behavior long tail and cold start problems
12
13 How to predict click behavior? Long tail and cold start problems
14 How to predict click behavior? Long tail and cold start problems
15 Long tail query demo: Google vs. Bing
16 A unified framework for click modeling Problem definition Definition 1: (Click modeling) Let random variable u denotes a user, q denotes a query issued by the user, a denotes an ad, r is the position of the ad. The binary variable c is 1 if the ad is clicked and 0 otherwise. Let L denotes the impression list and S denote the click sequence. Click modeling aims to explain observed click events. The shorthand is: P(c, q, a, u, r, L, S) Goals of click modeling 1 To estimate the actual ad relevance from biased clickthrough logs 2 To predict P(c = 1 q, a, u, r, L, S) for future impressions
17 An overview of click models Hypotheses in click modeling To model click events, we have to incorporate proper browsing hypotheses (i.e., generative process). The main hypotheses include: Unbiased hypothesis: P(c q, a, u, r, L, S) = P(c q, a) Position bias hypothesis: P(c q, a, u, r, L, S) = P(c q, a, r) Depend on click pattern: P(c q, a, u, r, L, S) = P(c q, a, r, S) : P(c q, a, u, r, L, S) = P(c q, a, r, L) Depend on user intent: P(c q, a, u, r, L, S) = P(c q, a, u, r)
18 Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
19 Unbiased hypothesis Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
20 Unbiased hypothesis Unbiased hypothesis: Basic hypothesis Basic hypothesis In the basic hypothesis, there is no bias associated with the observed clicks. This leads to the simplest model: P(c q, a, u, r, L, S) = P(c q, a) Remark In the basic hypothesis, the click probability is dominated by the relevance between query q and ad a.
21 Position bias hypothesis Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
22 Position bias hypothesis Position bias hypothesis: Examination hypothesis Examination hypothesis (WWW 07, Richardson et al.) Examination hypothesis assumes that an ad be clicked must be both examined (i.e. e = 1) and relevant: P(c = 1 q, a, u, r, L, S) =P(c = 1 q, a, r) Independence assumption = P(c = 1 e, q, a, r)p(e q, a, r) e {0,1} =P(c = 1 e = 1, q, a)p(e = 1 r) Examination hypothesis Novelty: The first attempt to model position bias
23 Position bias hypothesis Position bias hypothesis: Examination hypothesis Examination hypothesis (WWW 07, Richardson et al.) The position bias P(e = 1 r) can be experimentally measured by presenting users with the same ad at various positions on the page, and observing the user clicks. Remark In the examination hypothesis, the position bias is modeled with the queryindependent examination probability P(e r) and eliminated from the relevance estimation.
24 Depend on click pattern Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
25 Depend on click pattern Depend on click pattern: Cascade hypothesis Cascade hypothesis (WSDM 08, Carswell et al.) Cascade hypothesis assumes that an user scans each ad sequentially without any skips until she clicks on an ad and does not examine any additional ads after the click: P(e 1 = 1) = 1 P(e i = 1 e i 1 = 0) = 0 P(e i = 1 e i 1 = 1) = 1 c i 1 Novelty: The first attempt to model click pattern
26 Depend on click pattern Depend on click pattern: Cascade hypothesis Cascade hypothesis (WSDM 08, Carswell et al.) The probability of a click sequence with kth ad being clicked is: P(c = 1 r = k, q, a, u, L, S) =P(c = 1 r = k, q, a, L, S) Independence assumption k 1 =P(c = 1 r = k, q, a) P(c = 0 r = i, q, a) i=1 Cascade hypo. Remark This model is quite restrictive since it allows at most one click per query session.
27 Depend on click pattern Depend on click pattern: Multipleclick model MultipleClick Model (WSDM 09, Guo et al.) Novelty: To enable multiple clicks in a session by incorporating a decision phase for continuing examining results. Figure: The user model of dependent click model
28 Depend on click pattern Depend on click pattern: Multipleclick model Multipleclick model (WSDM 09, Guo et al.) The probability of examination and click is given by: P(e = 1 r = 1) = 1 P(c = 1 r = i) = P(e = 1 r = i)p(c = 1 e = 1, r = i) P(e = 1 r = i + 1) = λ i P(c = 1 r = i) + P(c = 0 r = i) The probability of a click sequence with kth ad being clicked is: P(c = 1 r = k, q, a, u, L, S) =P(c = 1 r = k, q, a, L, S) Independence assumption k 1 =P(c = 1 r = k, q, a) λ i P(c = 1 r = i, q, a) + P(c = 0 r = i, q, a) i=1
29 Depend on click pattern Depend on click pattern: Dynamic Bayesian Network Dynamic Bayesian Network (WWW 09, Chapelle and Zhang) Novelty: The first attempt to model post click pattern Key idea: Model both post and perceived relevance Figure: The DBN used for clicks modeling.
30 Depend on click pattern Depend on click pattern: Dynamic Bayesian Network Dynamic Bayesian Network (WWW 09, Chapelle and Zhang) The following equations describe the model: A i = 1, E i = 1 C i = 1 P(A i = 1) = a u P(S i = 1 C i = 1) = s u C i = 0 S i = 0 S i = 1 E i+1 = 0 P(E i+1 = 1 E i = 1, S i = 0) = γ E i = 0 E i+1 = 0 where γ is the probability that an user examines the next result if she is not satisfied with the current result.
31 Depend on click pattern Depend on click pattern: Dynamic Bayesian Network Experiment: Data: 58,000,000 sessions and 682,000 unique queries from the click logs of the UK market Xaxis: # of training sessions occurred at Position 1 Yaxis: MSE between the true CTRs and predicted CTRs
32 Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
33 : Temporal click model Temporal click model (SIGIR 09, Xu et al.) Key idea: (Externality) An ad may receive fewer clicks when codisplayed with high quality ads. Novelty: The first attempt to model ad externality Data study 1 Data: Ad impression sequences with exactly two ads. Two data sets are constructed by collecting one month (tens of millions) ads shown on north and south, respectively. Ground truth: Empirical CTR as the measure of ad quality. Experiment Setting: Group impressions with similar ad quality at Position 1 into one bin and plot the average CTR at Position 2, and vice versa.
34 : Temporal click model
35 : Temporal click model Data study 2 Experiment Setting: Group impressions with similar CTR at Position 1 in one bin and plot the percentage of events where the first click occurred at Position 2 and vice versa.
36 : Temporal click model Figure: The first click influenced by ad quality
37 : Temporal click model Temporal click model (SIGIR 09, Xu et al.) Positional rationality hypothesis: 1 Users examine both ads together to assess their qualities, 2 If the ad at Position 2 is much better than ad at Position 1, users would click the ad at Position 2 first
38 : Temporal click model The proposed method Input: clickthrough log of ad impression sequence A =< a 1, a 2 >. Output: the predicted CTR of ads. Generative process:
39 : Temporal click model Graphical model: E: examination variable, E {0, 1} R a : ad quality variable, R a [0, 1] U a : position bias variable, U a [0, 1] F: random variable for the first pick, F {a 1, a 2 } S: random variable for the repick, S {a 1, a 2 } C i : click random variable for ith click, C {0, 1}
40 : Temporal click model Experiments Data set: 0.3 million unique queries and 0.1 billion sessions shown at north. 1.1 million queries and 0.65 billion sessions shown at south. Evaluation: MSE between true CTRs and predicted CTRs. Baselines: 1) Naive CTR statistics (NS) estimates CTR by counting, and 2) Bayesian browsing model (BBM) (KDD 09 Liu et al.)
41 : Temporal click model
42 : Temporal click model Experiment results Both TCM and BBM are significantly better than NS for all query frequencies. TCM is noticeably better than BBM on lessfrequent queries but shows similar performance on frequent ones.
43 : Relational click prediction Relational click prediction (WSDM 12, Xiong et al.) Key idea: Click events would be influenced by the similarity between codisplayed Ads. Novelty: The first attempt to model similarity influence Figure: Two ad lists for query itunes account.
44 : Relational click prediction Data study Data: 0.7 million unique queries and 0.6 million unique ads from one month click logs Experiment setting: 1 Group ads into a specific context, i.e. a triple T =< q, a, r >, where q, a and r represent query, ad and position respectively. 2 Select a triple T that appears in multiple pageviews, i.e. l =< q, ad list >. 3 Calculate similarity between ad a and other ads in each l. 4 Compute empirical CTR of T in each l, and compare them with the average CTR of T on all pageviews. Evaluation: CTR T,l = CTR T,l CTR T CTR T
45 : Relational click prediction Xaxis denotes the similarity between a and other codisplayed ads. Yaxis denotes the average CTR l for different triples.
46 : Relational click prediction Data study CTR l is negatively correlated with the similarity between surrounding ads. The intuition is that: When the surrounding ads are similar to the given ad in their contents (or topics), it is likely that they will distract user s attention.
47 : Relational click prediction The proposed method Key idea: Modeling ads in an ad list together instead of treating them independently. (P(c q, a, u, L, r = 1),, P(c q, a, u, L, r = n)) T = F(X, R) where X = {x 1,, x n } includes all the feature vectors x i extracted from < q, a, u, L, r = i >, and R encodes the relation between ads.
48 : Relational click prediction Graphical model: Figure: A continuous CRF model for relational click prediction.
49 : Relational click prediction Let Y = {y 1,, y n } denotes the predicted CTRs of ads. The probability distribution of output Y conditioned on input X is defined as P(Y X) = 1 Z(X) exp h(y i, X; w) + βg(y i, y j, X) i j>i where h is the vertex feature function representing the dependence between CTR and input feature vectors, g is the edge feature function representing pairwise relationship between ads.
50 : Relational click prediction Individual modeling For simplicity, they define the vertex feature function as follows, h(y i, X; w) = (y i f (x i ; w)) 2 where f (x i ; w) is the output of any conventional click model. Relational modeling As discussed, if two ads are very similar to each other, their click probabilities will both become lower. To encode this intuition, they define the edge feature function as below. g(y i, y j, X) = s i,j (y i + y j ), where s i,j is the term similarity between ads i and j.
51 : Relational click prediction The whole model By combining all the feature functions, we obtain the overall conditional probability distribution: P(Y X) = 1 Z(X) exp (y i f (x i ; w)) 2 + βs i,j (y i + y j ) j>i i
52 : Relational click prediction Experiments Data set: 0.7 million unique queries and 0.6 million unique ads from one month click logs Extracted features: history COEC, relevance of ad to query, attractiveness of ad title and description, reputation of advertiser, etc. Baselines: Logistic Regression (LOCAL) and an variant of the proposed method with no edge features being used. Evaluation: MSE between true CTR and predicted CTR
53 : Relational click prediction The proposed method significantly outperform baselines. It performs better in lower positions than higher positions which consists with the cascade assumption. Figure: Results of NMSE.
54 : Relational click prediction They further study the performance of click prediction with respect to different levels of similarities in the ad lists. When the similarity between ads increases, the performance of CRF also increases. Figure: NMSE results at different similarity levels.
55 Depend on user intent Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
56 Depend on user intent Depend on user intent: Taskcentric click model Taskcentric click model (KDD 11, Zhang et al.) Novelty: The first attempt to model user behavior across multiple query sessions. Key ideas: Users tend to express their information needs incrementally, and click fresh documents that are not included before
57 Depend on user intent Depend on user intent: Taskcentric click model Taskcentric click model (KDD 11, Zhang et al.) Figure: The macro model of TCM.
58 Depend on user intent Depend on user intent: Taskcentric click model Taskcentric click model (KDD 11, Zhang et al.) Figure: The micro model of TCM.
59 Depend on user intent Depend on user intent: Taskcentric click model Graphical model:
60 Depend on user intent User intent demo
61 Outline 1 An overview of web search marketing 2 An overview of click modeling 3 A survey on click models Unbiased hypothesis Position bias hypothesis Depend on click pattern Depend on user intent 4 Future work
62 Future work Click modeling for long tail query, crowdsourcing? Automatic feature construction, deep learning? Evaluation metrics Click modeling for web search in mobile device very different user browsing behavior may be totally different business model
63 Reference WWW 07, Richardson et al.: Predicting clicks: estimating the clickthrough rate for new ads WSDM 08, Carswell et al.: An experimental comparison of click positionbias models WSDM 09, Guo et al.: Efficient multipleclick models in web search WWW 09, Chapelle and Zhang: A dynamic bayesian network click model for web search ranking SIGIR 09, Xu et al.: Temporal click model for sponsored search WSDM 12, Xiong et al.: Relational click prediction for sponsored search KDD 11, Zhang et al.: Userclick modeling for understanding and predicting searchbehavior
64 Thanks
Click Chain Model in Web Search
Click Chain Model in Web Search Fan Guo 1, Chao Liu, Anitha Kannan 3, Tom Minka 4, Michael Taylor 4, YiMin Wang, Christos Faloutsos 1 1 Carnegie Mellon University, Pittsburgh PA 1513, USA Microsoft Research
More informationInvited Applications Paper
Invited Applications Paper   Thore Graepel Joaquin Quiñonero Candela Thomas Borchert Ralf Herbrich Microsoft Research Ltd., 7 J J Thomson Avenue, Cambridge CB3 0FB, UK THOREG@MICROSOFT.COM JOAQUINC@MICROSOFT.COM
More information17.6.1 Introduction to Auction Design
CS787: Advanced Algorithms Topic: Sponsored Search Auction Design Presenter(s): Nilay, Srikrishna, Taedong 17.6.1 Introduction to Auction Design The Internet, which started of as a research project in
More informationNominal and ordinal logistic regression
Nominal and ordinal logistic regression April 26 Nominal and ordinal logistic regression Our goal for today is to briefly go over ways to extend the logistic regression model to the case where the outcome
More informationClick efficiency: a unified optimal ranking for online Ads and documents
DOI 10.1007/s1084401503663 Click efficiency: a unified optimal ranking for online Ads and documents Raju Balakrishnan 1 Subbarao Kambhampati 2 Received: 21 December 2013 / Revised: 23 March 2015 / Accepted:
More informationLearning to Rank Revisited: Our Progresses in New Algorithms and Tasks
The 4 th ChinaAustralia Database Workshop Melbourne, Australia Oct. 19, 2015 Learning to Rank Revisited: Our Progresses in New Algorithms and Tasks Jun Xu Institute of Computing Technology, Chinese Academy
More informationHow much can Behavioral Targeting Help Online Advertising? Jun Yan 1, Ning Liu 1, Gang Wang 1, Wen Zhang 2, Yun Jiang 3, Zheng Chen 1
WWW 29 MADRID! How much can Behavioral Targeting Help Online Advertising? Jun Yan, Ning Liu, Gang Wang, Wen Zhang 2, Yun Jiang 3, Zheng Chen Microsoft Research Asia Beijing, 8, China 2 Department of Automation
More informationPosition Auctions with Externalities
Position Auctions with Externalities Patrick Hummel 1 and R. Preston McAfee 2 1 Google Inc. phummel@google.com 2 Microsoft Corp. preston@mcafee.cc Abstract. This paper presents models for predicted clickthrough
More informationA Logistic Regression Approach to Ad Click Prediction
A Logistic Regression Approach to Ad Click Prediction Gouthami Kondakindi kondakin@usc.edu Satakshi Rana satakshr@usc.edu Aswin Rajkumar aswinraj@usc.edu Sai Kaushik Ponnekanti ponnekan@usc.edu Vinit Parakh
More informationOptimizing Search Engines using Clickthrough Data
Optimizing Search Engines using Clickthrough Data Presented by  Kajal Miyan Seminar Series, 891 Michigan state University *Slides adopted from presentations of Thorsten Joachims (author) and ShuiLung
More informationRevenue Optimization with Relevance Constraint in Sponsored Search
Revenue Optimization with Relevance Constraint in Sponsored Search Yunzhang Zhu Gang Wang Junli Yang Dakan Wang Jun Yan Zheng Chen Microsoft Resarch Asia, Beijing, China Department of Fundamental Science,
More informationBasics of Statistical Machine Learning
CS761 Spring 2013 Advanced Machine Learning Basics of Statistical Machine Learning Lecturer: Xiaojin Zhu jerryzhu@cs.wisc.edu Modern machine learning is rooted in statistics. You will find many familiar
More informationClickThrough Rate Estimation for Rare Events in Online Advertising
ClickThrough Rate Estimation for Rare Events in Online Advertising Xuerui Wang, Wei Li, Ying Cui, Ruofei (Bruce) Zhang, Jianchang Mao Yahoo! Labs, Silicon Valley United States ABSTRACT In online advertising
More informationStatistical Machine Learning
Statistical Machine Learning UoC Stats 37700, Winter quarter Lecture 4: classical linear and quadratic discriminants. 1 / 25 Linear separation For two classes in R d : simple idea: separate the classes
More informationData and Application
Carnegie Mellon University Data and Application Tutorial of Parameter Server Mu Li! CSD@CMU & IDL@Baidu! muli@cs.cmu.edu About me Ph.D student working with Alex Smola and Dave Andersen! large scale machine
More informationA General Framework for Mining ConceptDrifting Data Streams with Skewed Distributions
A General Framework for Mining ConceptDrifting Data Streams with Skewed Distributions Jing Gao Wei Fan Jiawei Han Philip S. Yu University of Illinois at UrbanaChampaign IBM T. J. Watson Research Center
More informationAn Empirical Analysis of Sponsored Search Performance in Search Engine Advertising. Anindya Ghose Sha Yang
An Empirical Analysis of Sponsored Search Performance in Search Engine Advertising Anindya Ghose Sha Yang Stern School of Business New York University Outline Background Research Question and Summary of
More informationIntroduction to Hypothesis Testing. Point estimation and confidence intervals are useful statistical inference procedures.
Introduction to Hypothesis Testing Point estimation and confidence intervals are useful statistical inference procedures. Another type of inference is used frequently used concerns tests of hypotheses.
More information8. Time Series and Prediction
8. Time Series and Prediction Definition: A time series is given by a sequence of the values of a variable observed at sequential points in time. e.g. daily maximum temperature, end of day share prices,
More informationConsiderations of Modeling in Keyword Bidding (Google:AdWords) Xiaoming Huo Georgia Institute of Technology August 8, 2012
Considerations of Modeling in Keyword Bidding (Google:AdWords) Xiaoming Huo Georgia Institute of Technology August 8, 2012 8/8/2012 1 Outline I. Problem Description II. Game theoretical aspect of the bidding
More informationModeling Contextual Factors of Click Rates
Modeling Contextual Factors of Click Rates Hila Becker Columbia University 500 W. 120th Street New York, NY 10027 Christopher Meek Microsoft Research 1 Microsoft Way Redmond, WA 98052 David Maxwell Chickering
More informationProbabilistic Models for Big Data. Alex Davies and Roger Frigola University of Cambridge 13th February 2014
Probabilistic Models for Big Data Alex Davies and Roger Frigola University of Cambridge 13th February 2014 The State of Big Data Why probabilistic models for Big Data? 1. If you don t have to worry about
More informationFinding Advertising Keywords on Web Pages. Contextual Ads 101
Finding Advertising Keywords on Web Pages Scott Wentau Yih Joshua Goodman Microsoft Research Vitor R. Carvalho Carnegie Mellon University Contextual Ads 101 Publisher s website Digital Camera Review The
More informationLecture 2: Introduction to belief (Bayesian) networks
Lecture 2: Introduction to belief (Bayesian) networks Conditional independence What is a belief network? Independence maps (Imaps) January 7, 2008 1 COMP526 Lecture 2 Recall from last time: Conditional
More informationPurchase Conversions and Attribution Modeling in Online Advertising: An Empirical Investigation
Purchase Conversions and Attribution Modeling in Online Advertising: An Empirical Investigation Author: TAHIR NISAR  Email: t.m.nisar@soton.ac.uk University: SOUTHAMPTON UNIVERSITY BUSINESS SCHOOL Track:
More informationClickthrough Prediction for Advertising in Twitter Timeline
Clickthrough Prediction for Advertising in Twitter Timeline Cheng Li 1, Yue Lu 2, Qiaozhu Mei 1, Dong Wang 2, Sandeep Pandey 2 1 School of Information, University of Michigan, Ann Arbor, MI, USA 2 Twitter
More informationWeb and Data Analysis. Minkoo Seo June 2015
Web and Data Analysis Minkoo Seo June 2015 About R user since 2011 Wrote this book Software Engineer at Google Korea http://mkseo.pe.kr/ These views are mine and mine alone and do not reflect the views
More informationInternet Advertising and the Generalized Second Price Auction:
Internet Advertising and the Generalized Second Price Auction: Selling Billions of Dollars Worth of Keywords Ben Edelman, Harvard Michael Ostrovsky, Stanford GSB Michael Schwarz, Yahoo! Research A Few
More informationOptimizing Display Advertisements Based on Historic User Trails
Optimizing Display Advertisements Based on Historic User Trails Neha Gupta, Udayan Sandeep Nawathe Khurana, Tak Yeon Lee Tumri Inc. Department of Computer San Mateo, CA Science snawathe@tumri.com University
More informationResponse prediction using collaborative filtering with hierarchies and sideinformation
Response prediction using collaborative filtering with hierarchies and sideinformation Aditya Krishna Menon 1 KrishnaPrasad Chitrapura 2 Sachin Garg 2 Deepak Agarwal 3 Nagaraj Kota 2 1 UC San Diego 2
More informationCell Phone based Activity Detection using Markov Logic Network
Cell Phone based Activity Detection using Markov Logic Network Somdeb Sarkhel sxs104721@utdallas.edu 1 Introduction Mobile devices are becoming increasingly sophisticated and the latest generation of smart
More informationSubordinating to the Majority: Factoid Question Answering over CQA Sites
Journal of Computational Information Systems 9: 16 (2013) 6409 6416 Available at http://www.jofcis.com Subordinating to the Majority: Factoid Question Answering over CQA Sites Xin LIAN, Xiaojie YUAN, Haiwei
More informationREVIEW ON QUERY CLUSTERING ALGORITHMS FOR SEARCH ENGINE OPTIMIZATION
Volume 2, Issue 2, February 2012 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: A REVIEW ON QUERY CLUSTERING
More informationHow to assess the risk of a large portfolio? How to estimate a large covariance matrix?
Chapter 3 Sparse Portfolio Allocation This chapter touches some practical aspects of portfolio allocation and risk assessment from a large pool of financial assets (e.g. stocks) How to assess the risk
More informationCollaborative Filtering. Radek Pelánek
Collaborative Filtering Radek Pelánek 2015 Collaborative Filtering assumption: users with similar taste in past will have similar taste in future requires only matrix of ratings applicable in many domains
More informationINTRODUCTION TO CODING THEORY: BASIC CODES AND SHANNON S THEOREM
INTRODUCTION TO CODING THEORY: BASIC CODES AND SHANNON S THEOREM SIDDHARTHA BISWAS Abstract. Coding theory originated in the late 1940 s and took its roots in engineering. However, it has developed and
More informationTechnical challenges in web advertising
Technical challenges in web advertising Andrei Broder Yahoo! Research 1 Disclaimer This talk presents the opinions of the author. It does not necessarily reflect the views of Yahoo! Inc. 2 Advertising
More informationGOOGLE ADWORDS. Optimizing Online Advertising. 15.071x The Analytics Edge
GOOGLE ADWORDS Optimizing Online Advertising 15.071x The Analytics Edge Google Inc. Provides products and services related to the Internet Mission: to organize the world s information and make it universally
More informationLinear Threshold Units
Linear Threshold Units w x hx (... w n x n w We assume that each feature x j and each weight w j is a real number (we will relax this later) We will study three different algorithms for learning linear
More informationVariance Reduction. Pricing American Options. Monte Carlo Option Pricing. Delta and Common Random Numbers
Variance Reduction The statistical efficiency of Monte Carlo simulation can be measured by the variance of its output If this variance can be lowered without changing the expected value, fewer replications
More informationContinuous Time Bayesian Networks for Inferring Users Presence and Activities with Extensions for Modeling and Evaluation
Continuous Time Bayesian Networks for Inferring Users Presence and Activities with Extensions for Modeling and Evaluation Uri Nodelman 1 Eric Horvitz Microsoft Research One Microsoft Way Redmond, WA 98052
More informationCSCI567 Machine Learning (Fall 2014)
CSCI567 Machine Learning (Fall 2014) Drs. Sha & Liu {feisha,yanliu.cs}@usc.edu September 22, 2014 Drs. Sha & Liu ({feisha,yanliu.cs}@usc.edu) CSCI567 Machine Learning (Fall 2014) September 22, 2014 1 /
More informationInternet Advertising and the Generalized SecondPrice Auction: Selling Billions of Dollars Worth of Keywords
Internet Advertising and the Generalized SecondPrice Auction: Selling Billions of Dollars Worth of Keywords by Benjamin Edelman, Michael Ostrovsky, and Michael Schwarz (EOS) presented by Scott Brinker
More informationPay Per Click Judo. Getting it Done Better and Faster by Doing What Works
Pay Per Click Judo Getting it Done Better and Faster by Doing What Works The Ancient Mysteries of PPC What do all those numbers in AdWords MEAN? How can I use them to create visitors of value? Leads, Sales
More information7. Tests of association and Linear Regression
7. Tests of association and Linear Regression In this chapter we consider 1. Tests of Association for 2 qualitative variables. 2. Measures of the strength of linear association between 2 quantitative variables.
More informationA Practical Application of Differential Privacy to Personalized Online Advertising
A Practical Application of Differential Privacy to Personalized Online Advertising Yehuda Lindell Eran Omri Department of Computer Science BarIlan University, Israel. lindell@cs.biu.ac.il,omrier@gmail.com
More informationDynamical Clustering of Personalized Web Search Results
Dynamical Clustering of Personalized Web Search Results Xuehua Shen CS Dept, UIUC xshen@cs.uiuc.edu Hong Cheng CS Dept, UIUC hcheng3@uiuc.edu Abstract Most current search engines present the user a ranked
More informationSponsored Search Ad Selection by Keyword Structure Analysis
Sponsored Search Ad Selection by Keyword Structure Analysis Kai Hui 1, Bin Gao 2,BenHe 1,andTiejianLuo 1 1 University of Chinese Academy of Sciences, Beijing, P.R. China huikai10@mails.ucas.ac.cn, {benhe,tjluo}@ucas.ac.cn
More informationPredictive Indexing for Fast Search
Predictive Indexing for Fast Search Sharad Goel Yahoo! Research New York, NY 10018 goel@yahooinc.com John Langford Yahoo! Research New York, NY 10018 jl@yahooinc.com Alex Strehl Yahoo! Research New York,
More informationCourse: Model, Learning, and Inference: Lecture 5
Course: Model, Learning, and Inference: Lecture 5 Alan Yuille Department of Statistics, UCLA Los Angeles, CA 90095 yuille@stat.ucla.edu Abstract Probability distributions on structured representation.
More informationPredict the Popularity of YouTube Videos Using Early View Data
000 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050
More informationThe ABCs of AdWords. The 49 PPC Terms You Need to Know to Be Successful. A publication of WordStream & Hanapin Marketing
The ABCs of AdWords The 49 PPC Terms You Need to Know to Be Successful A publication of WordStream & Hanapin Marketing The ABCs of AdWords The 49 PPC Terms You Need to Know to Be Successful Many individuals
More informationCompetitionBased Dynamic Pricing in Online Retailing
CompetitionBased Dynamic Pricing in Online Retailing Marshall Fisher The Wharton School, University of Pennsylvania, fisher@wharton.upenn.edu Santiago Gallino Tuck School of Business, Dartmouth College,
More information1 Maximum likelihood estimation
COS 424: Interacting with Data Lecturer: David Blei Lecture #4 Scribes: Wei Ho, Michael Ye February 14, 2008 1 Maximum likelihood estimation 1.1 MLE of a Bernoulli random variable (coin flips) Given N
More informationConstructing Social Intentional Corpora to Predict ClickThrough Rate for Search Advertising
Constructing Social Intentional Corpora to Predict ClickThrough Rate for Search Advertising YiTing Chen, HungYu Kao Department of Computer Science and Information Engineering National Cheng Kung University
More informationDecompose Error Rate into components, some of which can be measured on unlabeled data
BiasVariance Theory Decompose Error Rate into components, some of which can be measured on unlabeled data BiasVariance Decomposition for Regression BiasVariance Decomposition for Classification BiasVariance
More informationAn Introduction to Information Theory
An Introduction to Information Theory Carlton Downey November 12, 2013 INTRODUCTION Today s recitation will be an introduction to Information Theory Information theory studies the quantification of Information
More informationPoisson Models for Count Data
Chapter 4 Poisson Models for Count Data In this chapter we study loglinear models for count data under the assumption of a Poisson error structure. These models have many applications, not only to the
More informationSeven Stages of Pay Per Click Management
Seven Stages of Pay Per Click Management Optimising the performance of your PPC Campaign PPC Management Pay per click management is increasing complex. And pay per click market categories are increasingly
More informationInteraction between quantitative predictors
Interaction between quantitative predictors In a firstorder model like the ones we have discussed, the association between E(y) and a predictor x j does not depend on the value of the other predictors
More informationQuestion 2 Naïve Bayes (16 points)
Question 2 Naïve Bayes (16 points) About 2/3 of your email is spam so you downloaded an open source spam filter based on word occurrences that uses the Naive Bayes classifier. Assume you collected the
More informationBayes and Naïve Bayes. cs534machine Learning
Bayes and aïve Bayes cs534machine Learning Bayes Classifier Generative model learns Prediction is made by and where This is often referred to as the Bayes Classifier, because of the use of the Bayes rule
More informationThe Proportional Odds Model for Assessing Rater Agreement with Multiple Modalities
The Proportional Odds Model for Assessing Rater Agreement with Multiple Modalities Elizabeth GarrettMayer, PhD Assistant Professor Sidney Kimmel Comprehensive Cancer Center Johns Hopkins University 1
More informationHypothesis Testing. 1 Introduction. 2 Hypotheses. 2.1 Null and Alternative Hypotheses. 2.2 Simple vs. Composite. 2.3 OneSided and TwoSided Tests
Hypothesis Testing 1 Introduction This document is a simple tutorial on hypothesis testing. It presents the basic concepts and definitions as well as some frequently asked questions associated with hypothesis
More informationMaximum Likelihood Estimation
Math 541: Statistical Theory II Lecturer: Songfeng Zheng Maximum Likelihood Estimation 1 Maximum Likelihood Estimation Maximum likelihood is a relatively simple method of constructing an estimator for
More informationA Predictive Model for Advertiser ValuePerClick in Sponsored Search
A Predictive Model for Advertiser ValuePerClick in Sponsored Search Eric Sodomka Brown University Providence, Rhode Island Sébastien Lahaie Microsoft Research New York, New York Dustin Hillard Microsoft
More informationShould Ad Networks Bother Fighting Click Fraud? (Yes, They Should.)
Should Ad Networks Bother Fighting Click Fraud? (Yes, They Should.) Bobji Mungamuru Stanford University bobji@i.stanford.edu Stephen Weis Google sweis@google.com Hector GarciaMolina Stanford University
More informationα α λ α = = λ λ α ψ = = α α α λ λ ψ α = + β = > θ θ β > β β θ θ θ β θ β γ θ β = γ θ > β > γ θ β γ = θ β = θ β = θ β = β θ = β β θ = = = β β θ = + α α α α α = = λ λ λ λ λ λ λ = λ λ α α α α λ ψ + α =
More informationNormality Testing in Excel
Normality Testing in Excel By Mark Harmon Copyright 2011 Mark Harmon No part of this publication may be reproduced or distributed without the express permission of the author. mark@excelmasterseries.com
More informationLinear Classification. Volker Tresp Summer 2015
Linear Classification Volker Tresp Summer 2015 1 Classification Classification is the central task of pattern recognition Sensors supply information about an object: to which class do the object belong
More informationSibyl: a system for large scale machine learning
Sibyl: a system for large scale machine learning Tushar Chandra, Eugene Ie, Kenneth Goldman, Tomas Lloret Llinares, Jim McFadden, Fernando Pereira, Joshua Redstone, Tal Shaked, Yoram Singer Machine Learning
More information17. SIMPLE LINEAR REGRESSION II
17. SIMPLE LINEAR REGRESSION II The Model In linear regression analysis, we assume that the relationship between X and Y is linear. This does not mean, however, that Y can be perfectly predicted from X.
More informationArtificial Intelligence Mar 27, Bayesian Networks 1 P (T D)P (D) + P (T D)P ( D) =
Artificial Intelligence 15381 Mar 27, 2007 Bayesian Networks 1 Recap of last lecture Probability: precise representation of uncertainty Probability theory: optimal updating of knowledge based on new information
More informationAn Introduction to Time Series Regression
An Introduction to Time Series Regression Henry Thompson Auburn University An economic model suggests examining the effect of exogenous x t on endogenous y t with an exogenous control variable z t. In
More informationAgenda. Mathias Lanner Sas Institute. Predictive Modeling Applications. Predictive Modeling Training Data. Beslutsträd och andra prediktiva modeller
Agenda Introduktion till Prediktiva modeller Beslutsträd Beslutsträd och andra prediktiva modeller Mathias Lanner Sas Institute Pruning Regressioner Neurala Nätverk Utvärdering av modeller 2 Predictive
More informationMultiple Clicks Model for Web Search of Multiclickable Documents
Multiple Clicks Model for Web Search of Multiclickable Documents Léa Laporte, Sébastien Déjean, Josiane Mothe To cite this version: Léa Laporte, Sébastien Déjean, Josiane Mothe. Multiple Clicks Model
More informationCITY UNIVERSITY OF HONG KONG. Revenue Optimization in Internet Advertising Auctions
CITY UNIVERSITY OF HONG KONG l ½ŒA Revenue Optimization in Internet Advertising Auctions p ]zwû ÂÃÙz Submitted to Department of Computer Science õò AX in Partial Fulfillment of the Requirements for the
More informationChapter 9: Hypothesis Testing Sections
Chapter 9: Hypothesis Testing Sections 9.1 Problems of Testing Hypotheses Skip: 9.2 Testing Simple Hypotheses Skip: 9.3 Uniformly Most Powerful Tests Skip: 9.4 TwoSided Alternatives 9.5 The t Test 9.6
More informationComparing Tag Clouds, Term Histograms, and Term Lists for Enhancing Personalized Web Search
Comparing Tag Clouds, Term Histograms, and Term Lists for Enhancing Personalized Web Search Orland Hoeber and Hanze Liu Department of Computer Science, Memorial University St. John s, NL, Canada A1B 3X5
More informationLarge Scale Learning to Rank
Large Scale Learning to Rank D. Sculley Google, Inc. dsculley@google.com Abstract Pairwise learning to rank methods such as RankSVM give good performance, but suffer from the computational burden of optimizing
More informationData Mining in Web Search Engine Optimization and User Assisted Rank Results
Data Mining in Web Search Engine Optimization and User Assisted Rank Results Minky Jindal Institute of Technology and Management Gurgaon 122017, Haryana, India Nisha kharb Institute of Technology and Management
More informationThe Lane s Gifts v. Google Report
The Lane s Gifts v. Google Report By Alexander Tuzhilin Professor of Information Systems at the Stern School of Business at New York University, Report published July 2006 1 The Lane s Gifts case 2005
More informationDifferent Users and Intents: An Eyetracking Analysis of Web Search
Different Users and Intents: An Eyetracking Analysis of Web Search ABSTRACT Cristina GonzálezCaro Pompeu Fabra University Roc Boronat 138 Barcelona, Spain cgonzalc@unab.edu.co We present an eyetracking
More informationTests for Two Survival Curves Using Cox s Proportional Hazards Model
Chapter 730 Tests for Two Survival Curves Using Cox s Proportional Hazards Model Introduction A clinical trial is often employed to test the equality of survival distributions of two treatment groups.
More informationSocial Intelligence Report Adobe Digital Index Q2 2015
Social Intelligence Report Adobe Digital Index Q2 2015 Key Insights Paid Social Cost per click (CPC) rates for Facebook are flat YoY while impressions fell by half and click through rates doubled 51% of
More informationPearson's Correlation Tests
Chapter 800 Pearson's Correlation Tests Introduction The correlation coefficient, ρ (rho), is a popular statistic for describing the strength of the relationship between two variables. The correlation
More informationBALANCE LEARNING TO RANK IN BIG DATA. Guanqun Cao, Iftikhar Ahmad, Honglei Zhang, Weiyi Xie, Moncef Gabbouj. Tampere University of Technology, Finland
BALANCE LEARNING TO RANK IN BIG DATA Guanqun Cao, Iftikhar Ahmad, Honglei Zhang, Weiyi Xie, Moncef Gabbouj Tampere University of Technology, Finland {name.surname}@tut.fi ABSTRACT We propose a distributed
More informationSocial Media Mining. Network Measures
Klout Measures and Metrics 22 Why Do We Need Measures? Who are the central figures (influential individuals) in the network? What interaction patterns are common in friends? Who are the likeminded users
More informationSpatial Statistics Chapter 3 Basics of areal data and areal data modeling
Spatial Statistics Chapter 3 Basics of areal data and areal data modeling Recall areal data also known as lattice data are data Y (s), s D where D is a discrete index set. This usually corresponds to data
More informationA NURSING CARE PLAN RECOMMENDER SYSTEM USING A DATA MINING APPROACH
Proceedings of the 3 rd INFORMS Workshop on Data Mining and Health Informatics (DMHI 8) J. Li, D. Aleman, R. Sikora, eds. A NURSING CARE PLAN RECOMMENDER SYSTEM USING A DATA MINING APPROACH Lian Duan
More informationViewability Prediction for Online Display Ads
Viewability Prediction for Online Display Ads Chong Wang Information Systems New Jersey Institute of Technology Newark, NJ 07003, USA cw87@njit.edu Achir Kalra Forbes Media 499 Washington Blvd Jersey City,
More informationMATH4427 Notebook 2 Spring 2016. 2 MATH4427 Notebook 2 3. 2.1 Definitions and Examples... 3. 2.2 Performance Measures for Estimators...
MATH4427 Notebook 2 Spring 2016 prepared by Professor Jenny Baglivo c Copyright 20092016 by Jenny A. Baglivo. All Rights Reserved. Contents 2 MATH4427 Notebook 2 3 2.1 Definitions and Examples...................................
More informationWhy Taking This Course? Course Introduction, Descriptive Statistics and Data Visualization. Learning Goals. GENOME 560, Spring 2012
Why Taking This Course? Course Introduction, Descriptive Statistics and Data Visualization GENOME 560, Spring 2012 Data are interesting because they help us understand the world Genomics: Massive Amounts
More informationCollege Tuition: Data mining and analysis
CS105 College Tuition: Data mining and analysis By Jeanette Chu & Khiem Tran 4/28/2010 Introduction College tuition issues are steadily increasing every year. According to the college pricing trends report
More information4. Joint Distributions of Two Random Variables
4. Joint Distributions of Two Random Variables 4.1 Joint Distributions of Two Discrete Random Variables Suppose the discrete random variables X and Y have supports S X and S Y, respectively. The joint
More informationEstimating Sharer Reputation via Social Data Calibration
Estimating Sharer Reputation via Social Data Calibration Jaewon Yang Stanford University jayang@stanford.edu BeeChung Chen LinkedIn bchen@linkedin.com Deepak Agarwal LinkedIn dagarwal@linkedin.com ABSTRACT
More informationAn improved online algorithm for scheduling on two unrestrictive parallel batch processing machines
This is the PrePublished Version. An improved online algorithm for scheduling on two unrestrictive parallel batch processing machines Q.Q. Nong, T.C.E. Cheng, C.T. Ng Department of Mathematics, Ocean
More informationForecasting Trade Direction and Size of Future Contracts Using Deep Belief Network
Forecasting Trade Direction and Size of Future Contracts Using Deep Belief Network Anthony Lai (aslai), MK Li (lilemon), Foon Wang Pong (ppong) Abstract Algorithmic trading, high frequency trading (HFT)
More informationEFFECTIVE ONLINE ADVERTISING
EFFECTIVE ONLINE ADVERTISING by Hamed Sadeghi Neshat B.Sc., Sharif University of Technology, Tehran, Iran, 2009 a Thesis submitted in partial fulfillment of the requirements for the degree of Master of
More informationLearning to Predict SubjectLine Opens for LargeScale Email Marketing
Learning to Predict SubjectLine Opens for LargeScale Email Marketing Raju Balakrishnan Data Sciences, Groupon Inc. Palo Alto CA USA 94306 Email: raju@groupon.com. Rajesh Parekh Data Sciences, Groupon
More information