Presentation is loading. Please wait.

Presentation is loading. Please wait.

Xinyu Xing, Wei Meng, Dan Doozan, Georgia Institute of Technology Alex C. Snoeren, UC San Diego Nick Feamster, and Wenke Lee, Georgia Institute of Technology.

Published byDamian Hodges Modified over 9 years ago

Similar presentations

Presentation on theme: "Xinyu Xing, Wei Meng, Dan Doozan, Georgia Institute of Technology Alex C. Snoeren, UC San Diego Nick Feamster, and Wenke Lee, Georgia Institute of Technology."— Presentation transcript:

1 Xinyu Xing, Wei Meng, Dan Doozan, Georgia Institute of Technology Alex C. Snoeren, UC San Diego Nick Feamster, and Wenke Lee, Georgia Institute of Technology 22 nd USENIX Security (August, 2013)

2 Outline  Introduction  Overview and Attack Model  Pollution Attacks on YouTube  Google Personalized Search  Pollution Attacks on Amazon 2013/9/3A Seminar at Advanced Defense Lab2

3 Introduction  Modern Web services are increasingly relying upon personalization to improve the quality of their customers’ experience.  Many services with personalized content log their users’ Web activities. 2013/9/3A Seminar at Advanced Defense Lab3

4 This paper...  We demonstrate that contemporary personalization mechanisms are vulnerable to exploit. 2013/9/3A Seminar at Advanced Defense Lab4

5 Our Attack  We show that YouTube, Amazon, and Google are all vulnerable to the same class of cross-site scripting attack, which we call a pollution attack, that allows third parties to alter the customized content.  A distinguishing feature of our attack is that it does not exploit any vulnerability in the user’s Web browser. 2013/9/3A Seminar at Advanced Defense Lab5

6 Overview and Attack Model  The main instrument that a service provider can use to affect the content that a user sees is modifying the choice set.  When a user issues a query, a service’s personalization algorithm affects the user’s choice set for that query. 2013/9/3A Seminar at Advanced Defense Lab6

7 Overview and Attack Model (cont.)  In this paper, we focus on how changes to a user’s history can affect the choice set, holding other factors fixed.  This attack requires three steps: 1. Model the service’s personalization algorithm. 2. Create a “seed” to pollute the user’s history. 3. Inject the seed with a vector of false clicks. 2013/9/3A Seminar at Advanced Defense Lab7

8 2013/9/3A Seminar at Advanced Defense Lab8

9 Pollution Attacks on YouTube  Personalization rule Consider only those videos that the user watched for a long period of time Similar viewing histories Not recommend a video the user has already watched Two of suggested videos are recommended based upon personalization 2013/9/3A Seminar at Advanced Defense Lab9

10 2013/9/3A Seminar at Advanced Defense Lab10

11 Preparing Seed Videos 2013/9/3A Seminar at Advanced Defense Lab11 Video channel (C) ΩSΩS ΩTΩT

12 Inject Seed Videos  We see the video: http://www.youtube.com/user_watch? plid= &video_id=  We watch for a period of time: http://www.youtube.com/set_awesom e?plid= &video_id= 2013/9/3A Seminar at Advanced Defense Lab12

13 Experimental Design Relationship Account NewExisting New Two 3-minute videos (with about 65 sequentially watching) 100 channel (in top 2000) X 25 videos Existing (22 volunteers) Channel OnlyyouHappycamp X 15 videos 2013/9/3A Seminar at Advanced Defense Lab13

14 Evaluation  We evaluated the effectiveness of our pollution attacks by logging in as the victim user and viewing 114 representative videos. 2013/9/3A Seminar at Advanced Defense Lab14

15 Evaluation (New Accounts)  Successfully  we computed the Pearson correlation between the showing frequencies and the lengths of the target videos ○ 0.54 => medium the Pearson correlation between the showing frequencies and the view counts of the target videos ○ 0.23 => moderate 2013/9/3A Seminar at Advanced Defense Lab15

16 2013/9/3A Seminar at Advanced Defense Lab16

17 2013/9/3A Seminar at Advanced Defense Lab17

18 Evaluation (Existing Accounts)  For existing channel OnlyyouHappycamp 14 of the 22 volunteers (64%) Ten of our volunteers shared their histories The majority of the videos recommended to users for whom our attacks have low promotion rates have longer lengths and more view counts than our target videos. 2013/9/3A Seminar at Advanced Defense Lab18

19 2013/9/3A Seminar at Advanced Defense Lab19

20 Google Personalized Search  We describe two classes of personalization algorithms: contextual personalization persistent personalization 2013/9/3A Seminar at Advanced Defense Lab20

21 2013/9/3A Seminar at Advanced Defense Lab21

22 Identifying Search Terms  Contextual Personalization The keywords injected into a user’s search history should be both relevant to the promoting keyword and unique to the website being promoted. 2013/9/3A Seminar at Advanced Defense Lab22

23 Identifying Search Terms (cont.)  Persistent Personalization In this case, the size of the keyword set should be larger than that used for a contextual attack in order to have a greater effect on the user’s search history.  An attacker can safely inject roughly 50 keywords a minute using cross-site request forgery. we assume an attacker can inject at most 25 keywords into a user’s profile 2013/9/3A Seminar at Advanced Defense Lab23

24 Contextual Personalization 2013/9/3A Seminar at Advanced Defense Lab24 5,761 Search Terms from made-in-china.com 30 URLs URLs having unique keywords Google results 151,363 unique URLs 2,136 URLs 1,739 search terms

25 2,136 URLs for Contextual Personalization 2013/9/3A Seminar at Advanced Defense Lab25

26 Persistent Personalization 2013/9/3A Seminar at Advanced Defense Lab26 551 Search Terms from made-in-china.com 30 URLs URLs having unique Google AdWords keywords Google results 151,363 unique URLs 15,979 URLs

27 Evaluation  Contextual Personalization 2013/9/3A Seminar at Advanced Defense Lab27 1.1% 62.8% 28% 44%

28 Evaluation (cont.)  Persistent Personalization 2013/9/3A Seminar at Advanced Defense Lab28 4.3% 22.7% ??% 17%

29 Evaluation (cont.)  Real Users 97.1% of our 729 previously successful contextual attacks remain successful. Only 77.78% of the persistent pollution attacks that work on fresh accounts achieve similar success 2013/9/3A Seminar at Advanced Defense Lab29

30 Pollution Attacks on Amazon  Amazon tailors a customer’s homepage based on the previous purchase, browsing and searching behavior of the user.  We focused on the personalized recommendations Amazon generates based on the browsing and searching activities 2013/9/3A Seminar at Advanced Defense Lab30

31 2013/9/3A Seminar at Advanced Defense Lab31

32 Amazon Recommendations  Amazon’s personalization is based on history that maintained by the user’s web browser. Session cookie 2013/9/3A Seminar at Advanced Defense Lab32

33 Identifying Seed Products and Terms  Visit-Based Pollution the attacker visits the Amazon page of the product and retrieves the related products that are shown on Amazon page of the targeted product.  Search-Based Pollution An attacker could use a natural language toolkit to automatically extract a candidate keyword set from the targeted product’s name. 2013/9/3A Seminar at Advanced Defense Lab33

34 2013/9/3A Seminar at Advanced Defense Lab34

35 2013/9/3A Seminar at Advanced Defense Lab35

36 2013/9/3A Seminar at Advanced Defense Lab36

Download ppt "Xinyu Xing, Wei Meng, Dan Doozan, Georgia Institute of Technology Alex C. Snoeren, UC San Diego Nick Feamster, and Wenke Lee, Georgia Institute of Technology."

Similar presentations

Cross-Site Scripting Issues and Defenses Ed Skoudis Predictive Systems © 2002, Predictive Systems.

Cross-Site Scripting Issues and Defenses Ed Skoudis Predictive Systems © 2002, Predictive Systems.
Protecting Browser State from Web Privacy Attacks Collin Jackson, Andrew Bortz, Dan Boneh, John Mitchell Stanford University.

Protecting Browser State from Web Privacy Attacks Collin Jackson, Andrew Bortz, Dan Boneh, John Mitchell Stanford University.
Data Mining and Text Analytics Advertising Laura Quinn.

Data Mining and Text Analytics Advertising Laura Quinn.
Advanced Web Metrics with Google Analytics By: Carley Brown.

Advanced Web Metrics with Google Analytics By: Carley Brown.
Reporter: Jing Chiu Advisor: Yuh-Jye Lee 2015/7/181Data Mining & Machine Learning Lab.

Reporter: Jing Chiu Advisor: Yuh-Jye Lee /7/181Data Mining & Machine Learning Lab.
TI: An Efficient Indexing Mechanism for Real-Time Search on Tweets Chun Chen 1, Feng Li 2, Beng Chin Ooi 2, and Sai Wu 2 1 Zhejiang University, 2 National.

TI: An Efficient Indexing Mechanism for Real-Time Search on Tweets Chun Chen 1, Feng Li 2, Beng Chin Ooi 2, and Sai Wu 2 1 Zhejiang University, 2 National.
Center for E-Business Technology Seoul National University Seoul, Korea Socially Filtered Web Search: An approach using social bookmarking tags to personalize.

Center for E-Business Technology Seoul National University Seoul, Korea Socially Filtered Web Search: An approach using social bookmarking tags to personalize.
AUTOMATED DISCOVERY OF PARAMETER POLLUTION VULNERABILITIES IN WEB APPLICATIONS Marco Balduzzi, Carmen Torrano Gimenez, Davide Balzarotti, and Engin Kirda,

AUTOMATED DISCOVERY OF PARAMETER POLLUTION VULNERABILITIES IN WEB APPLICATIONS Marco Balduzzi, Carmen Torrano Gimenez, Davide Balzarotti, and Engin Kirda,
Design Guidelines for Effective WWW History Mechanisms Linda Tauscher and Saul Greenberg University of Calgary This talk accompanied a paper, and was presented.

Design Guidelines for Effective WWW History Mechanisms Linda Tauscher and Saul Greenberg University of Calgary This talk accompanied a paper, and was presented.
FACT: A Learning Based Web Query Processing System Hongjun Lu, Yanlei Diao Hong Kong U. of Science & Technology Songting Chen, Zengping Tian Fudan University.

FACT: A Learning Based Web Query Processing System Hongjun Lu, Yanlei Diao Hong Kong U. of Science & Technology Songting Chen, Zengping Tian Fudan University.
CONTENT-BASED BOOK RECOMMENDING USING LEARNING FOR TEXT CATEGORIZATION TRIVIKRAM BHAT UNIVERSITY OF TEXAS AT ARLINGTON DATA MINING CSE6362 BASED ON PAPER.

CONTENT-BASED BOOK RECOMMENDING USING LEARNING FOR TEXT CATEGORIZATION TRIVIKRAM BHAT UNIVERSITY OF TEXAS AT ARLINGTON DATA MINING CSE6362 BASED ON PAPER.
Personalized Ontologies for Web Search and Caching Susan Gauch Information and Telecommunications Technology Center Electrical Engineering and Computer.

Personalized Ontologies for Web Search and Caching Susan Gauch Information and Telecommunications Technology Center Electrical Engineering and Computer.
Lab 3 Cookie Stealing using XSS Kara James, Chelsea Collins, Trevor Norwood, David Johnson.

Lab 3 Cookie Stealing using XSS Kara James, Chelsea Collins, Trevor Norwood, David Johnson.
Copyright © 2007 - The OWASP Foundation Permission is granted to copy, distribute and/or modify this document under the terms of the Creative Commons Attribution-ShareAlike.

Copyright © The OWASP Foundation Permission is granted to copy, distribute and/or modify this document under the terms of the Creative Commons Attribution-ShareAlike.
WEB ANALYTICS Prof Sunil Wattal. Business questions How are people finding your website? What pages are the customers most interested in? Is your website.

WEB ANALYTICS Prof Sunil Wattal. Business questions How are people finding your website? What pages are the customers most interested in? Is your website.
1.Understand the decision-making process of consumer purchasing online. 2.Describe how companies are building one-to-one relationships with customers.

1.Understand the decision-making process of consumer purchasing online. 2.Describe how companies are building one-to-one relationships with customers.
AdWords Instructor: Dawn Rauscher. Quality Score in Action 0a2PVhPQhttp://www.youtube.com/watch?v=K7l 0a2PVhPQ.

AdWords Instructor: Dawn Rauscher. Quality Score in Action 0a2PVhPQhttp:// 0a2PVhPQ.
The 2nd International Conference of e-Learning and Distance Education, 21 to 23 February 2011, Riyadh, Saudi Arabia Prof. Dr. Torky Sultan Faculty of Computers.

The 2nd International Conference of e-Learning and Distance Education, 21 to 23 February 2011, Riyadh, Saudi Arabia Prof. Dr. Torky Sultan Faculty of Computers.
+ Websites Vulnerabilities. + Content Expand of The Internet Use of the Internet Examples Importance of the Internet How to find Security Vulnerabilities.

+ Websites Vulnerabilities. + Content Expand of The Internet Use of the Internet Examples Importance of the Internet How to find Security Vulnerabilities.
John P., Fang Yu, Yinglian Xie, Martin Abadi, Arvind Krishnamurthy University of California, Santa Cruz USENIX SECURITY SYMPOSIUM, August, 2010 John P.,

John P., Fang Yu, Yinglian Xie, Martin Abadi, Arvind Krishnamurthy University of California, Santa Cruz USENIX SECURITY SYMPOSIUM, August, 2010 John P.,

Similar presentations

Ads by Google