| Date |
2009-01-28 |
Time |
15:30:00 |
Room |
SH 107 |
| Speaker |
Kun Liu, IBM Almaden Research (Host: Giannella) |
| Abstract |
Social-networking sites have grown tremendously in popularity in recent years. Services such as Facebook and MySpace allow millions of users to create online profiles and share personal information with vast networks of friends, and often, strangers. As the number of users of these sites and the number of sites themselves explode, the privacy concerns surrounding both corporate and individual data take on increased importance.
In this talk, I will first present an overview of our very recent work on privacy-preserving social-network analysis in an effort to allow the audience to observe background and common themes. Then, I will introduce a systematic framework we have developed for identity anonymization on social
networks. This work considers the problem of sharing a social network owned by one corporate with another without revealing the identities of the registered users. Our framework strives to answer the following question: how to minimally modify the social-network graph to protect the identities
of the individuals? Finally, I will provide a set of recommendations for future research in this emerging area.
|
| Bio |
Kun Liu has been a postdoctoral researcher at IBM Almaden Research Center since January 2007. He received his Ph.D. from University of Maryland Baltimore County in May 2007. His research interests are in the area of privacy-preserving social-network analysis and text analytics for health care informatics. His work has won him two IBM Invention Achievement Awards and one IBM Bravo Award. Dr. Liu regularly serves on the program committee of many data mining conferences (e.g., KDD, ICDM, PKDD, PAKDD), and as a reviewer of many journals (e.g., IEEE TKDE, ACM TKDD). He has previously co-chaired the first SIAM international workshop on Practical Privacy-Preserving Data Mining. He is also maintaining the Privacy-preserving Data Mining (PPDM) Bibliography, one of the most popular PPDM paper repositories in the research community.
|
