Python anonymizer

Table of Contents

• The aim of this project
• Technologies
• The data used
• How this program works
• The list of recognised personal data
• Pseudonymisation techniques
• Anonymisation techniques

The aim of this project

The aim of this project is to develop an application which is able to detect personal data, and after detection anonymises and pseudonymises them.

Technologies

The project is written in Python 3.8. The dependencies can be found in requirements.txt.

The data used

• The data used in this project can be found in the data folder. The ds2.csv and test.csv files are used for testing purposes, they are generated, they DO NOT contain real data.
• The source of diseases_hungarian.csv in data/local folder is https://koronavirus.gov.hu/elhunytak
• The source of 21.02_disease_list.csv in data/local folder is https://www.targetvalidation.org/downloads/data

How this program works

The personal data detection functions can be found in detection.py. If the program detects personal data in a column, it saves the column name to labels.csv, and labels it based on the type of the data. Writing labels to a file is used in order to store the results for later use. The pseudonymisation.py module does the pseudonymisation and the anonymisation.py module does the anonymisation. The functions in datamanager.py make working with datasets easier. The analysis.py module can provide information to the user about the uniqueness of data in each column, and in the combination of columns. The auto_anon_and_pseud function in main.py automatizes the forementioned tasks.

The result (anonymised dataset) is written to data/output/outputtest.csv file.

The program includes a datacrawler package, which can be used for crawling data from different websites (currently just koronavirus.gov.hu).

The list of recognised personal data

• hungarian licence plates
• english disease names
• hungarian disease names
• hungarian tax number
• hungarian TAJ number
• hungarian personal number
• hungarian first names
• hungarian phone number
• MAC address
• IP address
• email address
• country codes and country names in english
• human age

Pseudonymisation techniques

• number to interval
• country name to region
• separating email address to three parts and hashing them
• text to number

Anonymisation techniques

K-anonymity, L-diversity, T-closeness. The implementation is based on https://github.com/Nuclearstar/K-Anonymity.