We recently presented a paper on Cantabular’s disclosure rules language at Eurostat’s conference on New Techniques and Technologies for Statistics.
The 10 minute presentation covered our innovative new coding language that helps statisticians automate decisions about table publication and was part of a session on input and output privacy in official statistics.
A shorter three minute version is shown below:
Disclosure rules language paper
When publishing population characteristic frequency tables for small geographic areas, such as in a census, it is usual that some cross-tabulations would be disclosive in some areas but not in others. Areas with higher diversity in person characteristics lead to more tables being non-disclosive. We describe a new computer language for codifying rules that decide whether an output table is disclosive for each geographic area. The publication rules are written in this language by the statistical disclosure control professionals and input into a flexible dissemination software system. The rules can thus be unknown to the programmers of the software system.
The full paper and a longer 10 minute video are available on the NTTS event website.
We’d like to thank Aleksandra Bujnowska from Eurostat for chairing the event.
Here’s a selection of other sessions from the conference we particularly enjoyed:
- Jörg Drechsler: Differential Privacy for Government Agencies – Are We There Yet?
- Fabian Bach: Differential Privacy and Noisy Confidentiality Concepts for European Population Statistics
- Jack Fitzsimons: Consideration for Privacy-Preserving Technologies for Official Statistics
Come to our event on the Irish 1911 census
1911 Irish census & technology: bringing the past, present and future together on 29th April 2021 at 14.30 BSTSign-up on Eventbrite