DG Update: May, 2001

National Institute of Statistical Sciences

Digital Government:

Project Update - May, 2001

1. RESEARCH. A functional prototype table server now exists, which is able, at realistic scale (currently, a 14-dimensional table with 300,000,000+ cells), to manipulate tables, calculate marginal subtables, and perform, from arbitrary sets of released marginals, both

(1) IPF reconstruction of the entire underlying table; and

(2) Computation of bounds for all non-zero entries in the underlying
table, using the "shuttle algorithm."

There is also a functioning query history database (QHDB) that currently tracks queries and the released frontier. The prototype was demonstrated at dg.o 2001 (see item 2 below). The next steps are to refine the QHDB and risk calculations; for example, so that when a table is considered for release the system will take into account which other tables would then become unreleasable.

2. PRESENTATIONS

May 22, 2001, at dg.o 2001 (Los Angeles)

Alan Karr and Ashish Sanil: Web-Based Systems that Disseminate
Information but Protect Confidential Data, including demonstration of
the prototype table server (see item 1 above).

Upcoming presentations include the ASA Spring Research Conference (Roanoke, June), the IMS New Researchers Conference (Atlanta, August), ISI 2001 (Seoul, August) and the 2001 FCSM Research Conference (Washington, November).

3. RECENT TECHNICAL REPORTS, all available on the project Web site, at
http://www.niss.org/dg/technicalreports.html:

Disclosure Limitation Methods and Information Loss for Tabular Data, George T. Duncan, Stephen E. Fienberg, Ramayya Krishnan, Rema Padman and Stephen F. Roehrig. To appear in Confidentiality, Disclosure and Data Access: Theory and Practical Applications for Statistical Agencies (sponsored by the Census Bureau and edited by Julia Lane)
U.S. Census Confidentiality: Perceptions and Reality, Stephen E. Fienberg and Margo Anderson. To appear in the Bulletin of the International Statistical Institute.
Web-Based Systems that Disseminate Information from Data but Protect Confidentiality, Alan F. Karr, Jaeyong Lee, Ashish P. Sanil, Joel Hernandez, Sousan Karimi and Karen Litwin. To appear in a volume on digital government edited by A. Elmagarmid and W. McIver and published by Kluwer.
Web-Based Systems that Disseminate Information but Protect Confidential Data, Alan F. Karr and Ashish P. Sanil. Proceedings of dg.o 2001.

4. OTHER ITEMS

Amanda Pyles, a 2001 graduate of Carnegie Mellon, will serve as an intern on the project this summer. She will work systematically on a few examples, including the Czech auto workers 6-way table and a 16-way table from the National Long Term Care Survey, to fit decomposable and graphical models, and to learn about the extent of disclosure risk when we release margins that are minimal sufficient statistics of reasonably good-fitting models.

Navigation: www.niss.org > Digital Government > Updates > May, 2001