Travel Grant Proposal for Workshop on Data Confidentiality

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

Abstract Proposal Number: 0741571 PI: Stephen E. Fienberg Institution: Carnegie Mellon University Title: Travel Grant Proposal for Workshop on Data Confidentiality Abstract Carnegie Mellon University will host a workshop September 6-7, 2007 on data confidentiality. Increasingly, organizations are collecting data to, among other things, be made available to researchers. Many kinds of data are needed by researchers, including: - Census data - Health data - Network data, for example collected from users' access to the Internet or general background data collected from routers or other network services. Building on his extensive background involving security for statistical databases, the PI has assembled an impressive collection of experts. This workshop will explore ways to apply existing statistical methods for sanitization of data for these applications and, as needed, new methods suitable to other applications. Before data can be released to researchers it is essential that data confidentiality policies be respected. For example, from a released data set associated with the Census Bureau, it should not be possible to associate a particular individual or household with particular sensitive data items, such as household income. Network data made available should not associate a packet, and its associated destination on the Internet or Web, with an individual or even an IP address. Various techniques will be considered for the specification of data confidentiality requirements and for sanitization of the data to meet the requirements, including masking fields and data transformation techniques that preserve properties essential for the research use of the data. The technical topics to be covered in the workshop are as follows: - Languages to express confidentiality needs - Legal, regulation and policy issues - Societal and economic impacts - Limitations of the technology, particularly in the face of the limitations of cryptographic methods - Applications, including network data, census data, other databases - Education and training - Needed infrastructure support The PI will produce a report on the workshop's deliberations. The proposed workshop will assemble Government, industry and academic organizations representing different disciplines with diverse needs for data, and will consider different techniques for data sanitization. Microsoft and IBM will be co-sponsoring the workshop. Microsoft's needs, for example, are to sanitize data gathered from use of their search engines and browsers.