Project Description

Social-networking sites (e.g., Facebook, MySpace, LinkedIn, etc.) and other online collaborative tools have emerged as places where people can post and share information.  This information-sharing has many benefits, ranging from practical  (e.g., sharing a business document) to purely social (e.g., communicating with distant friends). At the same time, information sharing inevitably poses significant threats to user privacy. In social-networking sites, for example, documented threats range from identity theft to digital stalking and personalized spam.  As a result, a growing number of such sites allow individual users to specify fine-grained policies that indicate who can access their data, and to what extent. However, studies have consistently shown that most end-users find the task of specifying access-control policies for their own data overwhelming; as a result, users often skip the process altogether.

This project is a collaboration between investigators at the University of Michigan and Boston University.  The goal of the project is to help social-media users gain control of their data.  To that end, the project includes three main components: assisted specification, feedback, and refinement recommendations.  To assist users in initially specifying access-control policies for their data, we are developing a "privacy wizard," which employs data mining and machine learning methods, including active learning, to construct an accurate policy, with minimal input from the user.  To provide feedback regarding existing privacy settings, we consider two approaches: aggregate numeric scores and visualizations.   Finally, we consider how numeric scores and visual feedback can be enriched with recommendations for refinements to help the user achieve a desired level of exposure.

Project Team

Kristen LeFevre (Michigan PI)

Evimaria Terzi (BU PI)

Lujun Fang (Michigan Ph.D. student)

Alessandra Mazzia (Michigan Ph.D. student)

Xuan Zhang (BU Ph.D. student)

Rajesh Bejugam (Michigan MS student)

Heedo Kim (Michigan BS/MS student)

Aaron Tami (Michigan undergraduate)

Collaborators: Eytan Adar (Michigan), Li Qian (Michigan), Kun Liu (Yahoo!)

Publications

Alessandra Mazzia, Kristen LeFevre, and Eytan Adar. A Tool for Privacy Comprehension.  CHI Workshops, 2011.

Lujun Fang, Heedo Kim, Kristen LeFevre, and Aaron Tami.  A Privacy-Recommendation Wizard for Users of Social Networking Sites.  Demonstration in CCS, 2010. 

Lujun Fang and Kristen LeFevre.  Privacy Wizards for Social Networking Sites.  WWW, 2010. Best Student Paper.

Kristen LeFevre and Evimaria Terzi.  GraSS: Graph Structure Summarization.  SIAM Data Mining, 2010.

Kun Liu, Evimaria Terzi: A framework for computing the privacy score of users in online social networks. ACM Transactions on  Knowledge Discovery from Data.  (To appear)

Kenneth Clarkson, Kun Liu, Evimaria Terzi: Towards Identity Anonymization in Social Networks. Book Chapter in Link Mining: Models Algorithms and Applications. Editors: C. Faloutsos, J. Han and P. Yu. (To appear)

Kun Liu, Evimaria Terzi: A framework for computing the privacy score of users in online social networks. ICDM, 2009.

Kun Liu, Evimaria Terzi: Towards identity anonymization on graphs. SIGMOD, 2008.

Support

This project is supported by National Science Foundation Grants CNS-10170149 and CNS-1017529.