Open Anonymizer is a natural language processing software that anonymizes German texts. Open Anonymizer uses a XLM-RoBERTa model for anonymizing persons/names, locations, organizations and a rule-based approach for anonymizing dates, phone numbers, other numbers and emails.
We also offer a graphical interface that you can see and start as described below. You can try the web-application via https://openanonymizer.codes/.
Make sure that poetry is installed on your machine (https://python-poetry.org/docs/#installation) and within this directory initialize your virtual environment with poetry shell afterwards you can start the application via
streamlit run app.py
This might take a while to start as the model is loaded when first used.
You can anonymize single texts
or use the file upload
This application can also be viewed here https://openanonymizer.codes/
You can also use Docker to run this application on another computer of your choice. However if you decide to do so, we strongly suggest that you download the model before and run the application with a production flag as described below.
poetry shell
poetry install
python3 get_model.py
docker build . -t open-anonymizer-webapp
docker run -e DEPLOYMENT=true -p 8501:8501 open-anonymizer-webapp
Open Anonymizer is a part of our Master Thesis for the Mannheim Master of Applied Data Science & Measurement written by Dimitri Epp and Markus Nutz.
