Open Diffix

Open source tools for anonymizing structured data

Open Diffix is an open source project developing tools for anonymizing structured data. The underlying technology for Open Diffix tools was developed by the Max Planck Institute for Software Systems in partnership with the former Aircloak GmbH.

SynDiffix

Our flagship tool is SynDiffix, a Python package for generating statistically accurate and strongly anonymous synthetic data from structured data. SynDiffix is:

• one to two orders of magnitude more accurate than other open-source tools, and
• five to ten times more accurate than the best commercial products.

SynDiffix is particularly well suited to descriptive analytics use cases. Nevertheless, it works for ML use cases as well.

The current release is the first release. We are planning many new features. If you would like to request features, or need help or advice, you may contact us at hello@open-diffix.org. You may also post bug reports or feature requests as GitHub issues at github.com/diffix/syndiffix.

pg_diffix

pg_diffix is a PostgreSQL extension that gives aggregate anonymized answers to SQL queries. Advantages over SynDiffix include:

• Faster and more scalable
• Supports untrusted users
• Supports back-end data for applications

pg_diffix only returns aggregate answers (column values and associated counts) to a very limited subset of SQL. It is far less flexible than SynDiffix. See github.com/diffix/pg_diffix.

Development on pg_diffix is currently paused. If you are interested in a project using pg_diffix, please contact us at hello@open-diffix.org.