Skip to content

The Flow Project

This is the documentation of the Flow Project, which is a collaborative research initiative (2023–2026) funded by the Swiss National Science Foundation (SNF) and the German Research Foundation (DFG).

Flow-Logo

What is Flow?

Flow develops standardized, user-friendly digital workflows for historians and humanities researchers working with premodern historical sources. The goal is to make machine-learning capabilities accessible without requiring data science expertise, enabling corpus-level analysis of extended document collections.

What can you do with it?

  • Preprocessing Prepare your data for further HTR tasks
  • Handwritten Text Recognition (HTR) Transcribe historical manuscripts automatically
  • Data Management Manage your existing data via HuggingFace Hub

How is it structured?

The Flow environment is modular. It consists of independent packages (Python libraries) and services (FastAPI microservices) that cover the HTR/ATR pipeline: preprocessing, inference, and evaluation. These are orchestrated through an n8n automation environment, so the pipeline can be run through web forms without writing code.

Where to start

Partners

Developed jointly by the University of Bern (Digital Humanities), the University of Bielefeld (Digital History), and the Research Centre for Hanse and Baltic History, Lübeck.

University of Bern - Digital Humanities University of Bielefeld - Digital History Research Centre for Hanse and Baltic History

License

This documentation, and the packages and services that make up the Flow environment, are released under the MIT license. n8n itself is licensed separately, under its own Sustainable Use License; see the FAQ for what that means in practice.

Citation

A scientific paper describing the Flow Project is in preparation and has not been published yet; this section will be updated with the full citation once it is.