anjana
Data anonymization
Apply nine anonymmization techniques, from k-anonymity, l-diversity, t-closeness to β-likeness and δ-disclosure privacy among others to tabular datasets. Designed for privacy-compliant data sharing in research and industry.
IFCA Privacy Toolbox
Privacy-preserving tools for handling sensitive data
A suite of open-source Python libraries for data anonymization, local differential privacy, and privacy risk assessment. Developed at IFCA for researchers and data practitioners.
pip install anjana pyCANON trasgoDP The tools
Three libraries, a complete privacy-preserving workflow
Each tool covers a different stage of the privacy pipeline. Use them independently or as a complete suite.
pyCANON
Privacy risk assessment
Check and measure the level of anonymization of a tabular dataset. Compute the parameters related to the nine anonymziation techniques implemented in anjana. In addition, provides metrics for checking the privacy-utiloty trde-offs and the risks of re-identification and disclousre. It allows to audit your data before release.
trasgoDP
Local differential privacy and metric privacy
Apply local based-differential privacy mechanisms directlly to tabular, both numerical, categorical and location-based data (via geo-indistiguisability). It implementes a correlation loss metric and allwos to check the divergence. Designed for realeasing sanitized versions of the data with the privacy guarantees provided by the privacy budget.
How it fits together
A complete privacy pipeline
The three tools are designed to work independently, but in a complementary way, covering the full lifecycle of privacy-preserving data publishing.
01 Anonymize or sanitize Apply anonymization techniques to tabular data using
anjana or sanitize the data using local-differential privacy mechanisms or metric-privacy notions with trasgoDP 02 Verify the privacy level Differential privacy (DP) provides mathematical guarantees on the level of privacy according to the privacy budget, and the level of anonymity can be check with
pyCANON 03 Check risks and utility Check privacy-utility trade-off metrics, re-identification and disclosure risks using
pyCANON for anonymized data. For data sanitized with trasgoDP check divergences and correlation loss Standards-based
Implements well-known privacy models (from anonymity to DP) with a strong theoretical background.
Open source
All code is available on GitHub under Apache 2.0. Auditable and reproducible.
Well documented
Each tool has its own ReadTheDocs documentation with starting guideliness and usage examples.
Research-grade
Built by researchers for researchers. Presented in peer-reviewed publications in high impact journals.
Used in
Research projects
These tools are actively used in scientific projects:
EOSC SIESTA
EU Horizon EuropeSecure Interactive Environments for SensiTive data Analytics. The SIESTA platform provides access to the interactive web application of the three tools, and these are also integrated in the SIESTA dashboard.
anjanapyCANONtrasgoDP
AI4EOSC
EU Horizon EuropeArtificial Intelligence for the European Open Science Cloud. pyCANON was exploited for analyzing the impact of anonymization in the performance of machine learning models.
pyCANON
EOSC ARENA
EU Horizon EuropeAI Research Enhancement through Networked Agents. This toolkit will be integrated in the AI4EOSC catalog by EOSC ARENA for dedicated training in responsible use of AI and data privacy.
anjanapyCANONtrasgoDP
Publications
Related papers
Peer-reviewed publications presenting the tools, methods and algorithms by the development team.
Scientific Data, vol. 11, no. 1, p. 1289, 2024
Scientific Data, vol. 9, no. 1, p. 785, 2022
IEEE International Conference on Cyber Security and Resilience (CSR), Venice, 2023
IEEE Xplore, 2nd International Conference on Federated Learning and Intelligent Computing Systems (FLICS), 2026
arXiv preprint, 2026
Papers exploiting the tools
Selected works by collaborators and associated projects that use the tools (for more third-party citations, refer to Google Scholar).
28th European Conference on Artificial Intelligence, Oct 2025, Bologna, Italy. pp.1009 - 1018, 2025
Preprint, 2025
Ready to protect your data?
Install the tools, read the docs, or reach out to the team.
About
Developed at IFCA
IFCA Privacy Toolbox is developed and maintained by the Advanced Computing and e-Science Group at the Institute of Physics of Cantabria (IFCA) , a joint centre of CSIC and the University of Cantabria.
If you use these tools in your research, please cite the corresponding publications. See individual documentation pages for citation details.
