Architecture

Project blackbar contains tooling allowing to anonymize and pseudonymize texts

• Push text data to annotate to Inception
• Extract annotated texts from Inception
• Build models which identify entities in order to anonymize text records
• Automated anonymization of text records
• Functionalities for pseudonymization of anonymized texts records
• Automated pseudonymization of anonymized text records
• Apps for validation, checks, model building and automation
• AI Apps on top of the pseudonymized texts

High level infrastructure

• Database with texts: IRIS or an ANSI SQL database
• Inception for annotation & generating training data
• NLP models (spacy / pytorch / transformers), alignment: Smith-Waterman
• Automation: Prefect
• Model storage backend: Minio
• Docker/Podman-based tooling
• Python packages
• Webapps behind ShinyProxy

The project consists of the following repositories

• blackbar with source code at https://github.com/bnosac/blackbar
• blackbar-py with source code at https://github.com/bnosac/blackbar-py
• blackbar-docker
• textalignment with source code at https://github.com/bnosac/textalignment
• rlike with source code at https://github.com/bnosac/rlike

The documentation of these are public, to have access to the source code, contact us here.