R E A D - C O O P

Loading

Gothic scripts from the Middle Ages can be found in archives and libraries all over Europe.  The script was widely used for hundreds of years, and not only in expensive decorated books.  First experiments with documents from Switzerland and Germany have demonstrated that Gothic script can be recognised by Automated Text Recognition models with good levels of accuracy (see an example from the cartulary of the Königsfelden abbey).

The next step is to combine different examples of Gothic scripts in order to build and improve generic models for the recognition of this kind of document.  Dr Tobias Hodel (State Archives of Zurich, University of Zurich) has set up the ‘Gothic Hands’ working group – where all Transkribus users can work together towards the aim of the improved recognition of Gothic material.  Scroll down to find out more about joining the working group and its aims.

St. Gallen, Stiftsbibliothek, Cod. Sang. 857, p. 124.  The St. Gall Nibelung manuscript B with the Nibelungenlied (The Song of the Nibelungs) and “Klage” (lament), “Parzival” and “Willehalm” by Wolfram von Eschenbach, and Stricker’s “Karl der Grosse” (Charlemagne) (https://www.e-codices.ch/en/list/one/csg/0857).

The process of combining training data of different Gothic documents has already been started as part of a collaboration between various digital editions projects (Parzival editionKönigsfelden edition and the Ortsbürgerarchiv St. Gallen, Konventsbuch St. Katharinental edition).  The resulting model (Comb_Gothic_Bookwriting) is now available to all Transkribus users – if you work with Gothic script, try it out!  The model can already transcribe Gothic documents with a Character Error Rate of less than 10%.  But this could be just the beginning!

The ‘Gothic Hands’ working group is looking for further examples of documents written in Gothic scripts from the 13th, 14th, and 15th century.  You can help us add to the collection – all that is needed are images and transcriptions. You can:

  • share existing training data that you have already prepared in Transkribus
  • prepare new images and transcripts in Transkribus in the ‘Gothic Hands’ collection
  • send over files containing images and transcripts which can be matched automatically and converted into training data using our Text2image tool.

To join the working group and get access to the ‘Gothic Hands’ collection in Transkribus, contact Tobias Hodel (tobias.hodel@hist.uzh.ch).

The ‘Gothic Hands’ working group aims to demonstrate that training based algorithms like Automated Text Recognition need significant input from many stakeholders – they can only be improved by cooperation and sharing!  This way of thinking aligns perfectly with the future of the READ project.  After the end of our European Union funding in June 2019, READ will become a European Cooperative Society (SCE) run for the benefit of its members.  More information about this new direction will be shared soon on the READ website and during the upcoming Transkribus User Conference in Vienna this November.