Anonymization and Uncertainty in Social Network Data
Abstract
Many privacy questions arise when collecting information
about individuals: how to allow use of this data without compromising the
privacy of the data subjects? How to ensure that the end users of this
data can find useful answers to their queries? Various anonymization
techniques have been proposed which aim to find meaningful tradeoffs
between privacy and utility. This talk presents the background for
privacy and anonymization. Then I'll describe methods for anonymizing
social network data, represented as a large, sparse semantic graph, which
is additionally challenging due to the complex patterns of interactions
between individuals. I'll also discuss some unexpected connections to
uncertain data management, which provides many further directions for
future work.
This talk covers join work with Divesh Srivastava, Balachander
Krishnamuthy, and Smriti Bhagat.
|
