pycanon 1.2.1

pyCANON: A Python library to check the level of anonymity of a dataset.

pyCANON is a Python library and CLI to assess the values of the parameters associated with the most common privacy-preserving techniques via anonymization.

Authors: Judith Sáinz-Pardo Díaz and Álvaro López García (IFCA - CSIC).

Installation

We recommend to use Python3 with virtualenv:

virtualenv .venv -p python3
source .venv/bin/activate

Then run the following command to install the library and all its requirements:

pip install pycanon

If you also want to install the functionality that allows to generate PDF files for the reports, install as follows

pip install pycanon[PDF]

Documentation

The pyCANON documentation is hosted on Read the Docs.

Getting started

Example using the adult dataset:

import pandas as pd
from pycanon import anonymity, report

FILE_NAME = "adult.csv"
QI = ["age", "education", "occupation", "relationship", "sex", "native-country"]
SA = ["salary-class"]
DATA = pd.read_csv(FILE_NAME)

# Calculate k for k-anonymity:
k = anonymity.k_anonymity(DATA, QI)

# Print the anonymity report:
report.print_report(DATA, QI, SA)

Description

pyCANON allows to check if the following privacy-preserving techniques are verified and the value of the parameters associated with each of them.

Technique	pyCANON function	Parameters	Notes
k-anonymity	k_anonymity	k: int
(α, k)-anonymity	alpha_k_anonymity	α: float k:int
ℓ-diversity	l_diversity	ℓ: int
Entropy ℓ-diversity	entropy_l_diversity	ℓ: int
Recursive (c,ℓ)-diversity	recursive_c_l_diversity	c: int ℓ: int	Not calculated if ℓ=1
Basic β-likeness	basic_beta_likeness	β: float
Enhanced β-likeness	enhanced_beta_likeness	β: float
t-closeness	t_closeness	t: float	For numerical attributes the definition of the EMD (one-dimensional Earth Mover’s Distance) is used. For categorical attributes, the metric “Equal Distance” is used.
δ-disclosure privacy	delta_disclosure	δ: float

More information can be found in this paper.

In addition, a report can be obtained including information on the equivalence clases and the usefulness of the data. In particular, for the latter the following three classically used metrics are implemented (as defined in the documentation):

• average equivalence class size

• classification metric

• discernability metric

Citation

If you are using pyCANON you can cite it as follows:

@article{sainzpardo2022pycanon,

title={A Python library to check the level of anonymity of a dataset}, author={S{'a}inz-Pardo D{'i}az, Judith and L{'o}pez Garc{'i}a, {'A}lvaro}, journal={Scientific Data}, volume={9}, number={1}, pages={785}, year={2022}, publisher={Nature Publishing Group UK London}}

Related work

If you are using pyCANON, you may also be interested in:

• anjana: a Python library for anonymizing tabular datasets.

• trasgoDP: a Python library which implements different mechanisms to apply local differential privacy directly to your data.

Acknowledgments

The authors would like to thank the funding through the European Union - NextGenerationEU (Regulation EU 2020/2094), through CSIC’s Global Health Platform (PTI+ Salud Global) and the support from the projects AI4EOSC “Artificial Intelligence for the European Open Science Cloud” (funded by the European Union’s Horizon Europe research and innovation programme under grant agreement number 101058593) and EOSC SIESTA (funded by European Union’s Horizon Europe research and innovation programme under grant agreement number 101131957).