EOSC-SIESTA develops tools and services to make the secure sharing and analysis of sensitive data more feasible within the EOSC ecosystem. The project addresses a key challenge in open science: enabling access to valuable datasets while preserving privacy and confidentiality. To support this objective, EOSC-SIESTA is building practical, user-oriented solutions that help researchers and institutions work with sensitive data in trusted environments, combining usability with privacy-preserving approaches.
Protecting sensitive data is not only about restricting access. It also requires understanding the level of privacy a dataset actually provides before it is shared or reused. This is where pyCANON comes in.
Developed within the EOSC-SIESTA ecosystem, pyCANON is a Python library and command-line interface designed to assess the anonymity of tabular datasets. It helps users evaluate privacy risks by applying a wide range of privacy-preserving models and generating detailed reports that support informed decisions on data sharing and reuse.
The tool implements several well-known anonymity models, including k-anonymity, (α,k)-anonymity, ℓ-diversity, t-closeness, β-likeness and δ-disclosure privacy. Thanks to this broad coverage, pyCANON can be useful for different kinds of users, from data scientists and researchers to institutions and policymakers working with sensitive information.
Within EOSC-SIESTA, pyCANON is also available through a web application, making it easier to explore its functionality in a more accessible way. Through the EOSC-SIESTA dashboard, users can launch the service, upload a dataset, define identifiers, quasi-identifiers and sensitive attributes, and then evaluate the dataset against one or several anonymity techniques. The results are displayed on screen and can also be downloaded in JSON format for further analysis or record-keeping.
To help users get started, EOSC-SIESTA has prepared some video tutorials showing how pyCANON works in practice. The tutorial walks through the full process, from accessing the service through the dashboard to uploading a dataset and interpreting the anonymity assessment results. It also shows how the tool can be used to compare a non-anonymised dataset with a previously anonymised version, illustrating how privacy levels change depending on the treatment applied to the data.
This makes pyCANON not only a useful assessment tool, but also a practical resource for understanding anonymity metrics in real workflows involving sensitive data.To learn more about pyCANON, you can explore its GitHub repository, read the documentation, consult the Scientific Data paper, and watch the tutorial available through EOSC-SIESTA.