Optimal tabular releases from confidential data
Alan F. Karr, Adrian Dobra, Ashish P. Sanil · International Conference on Digital Government Research · 2002
We describe and illustrate NISS-developed optimal tabular release technology, which releases sets of sub-tables of large contingency tables that maximize data utility (in our examples, the number of sub-tables released) subject to a constraint on disclosure risk (tightness of bounds on small-count, risky cells in the underlying table). This approach explicitly accommodates the mandate of Federal statistical agencies to protect data confidentiality and their mission to disseminate information derived from the data.