Workflows For the Humanities: Transkribus / Topic Modelling

The PHAIDRA department is building digital workflows around the transcription platform Transkribus, which accompanies the data cycle from the digitisation of sources to long-term archiving. Our goal is to make the data available in high quality according to the FAIR criteria for research and teaching at the University of Vienna.

Transkribus Team Univie:

Management: Mag. Martin Gasteiner
Training, Practice, User Management: Victoria Eisenheld, BA und Mag. Martin Gasteiner
Technical questions: Dr. Janos Bekesi
Scientific advice: Dr. Thomas Wallnig



Concrete service offers:

  • Credit allocation for the Transkribus system. Credits will be activated after application. (Application: informally in an email with the university email address or student address /500 credits are freely available when entering Transkribus/ and the scope of the transcription project).
  • Offer: Introduction to Transkribus or fixed consultation appointments. Bookable at:
  • Topic Modelling for large data sets. (Topic Modeling: Topics are recognised on the basis of statistical models in large amounts of text). An example can be found here: If you are interested, please contact

Short description of the services

What is Transkribus?

Based on neural network technology, the programme creates transcriptions of handwritten or printed material based on uploaded scans. In recent years, there have been some initiatives at our university to share experiences in this regard (cf. [p. 246). Introductions or experiments with your source are done as individual consultations.

Introductions and FAQs on Transkribus at:

Transkribus at the University of Vienna

In 2019, the University of Vienna joined the Transkribus - Consortium and thus has discounted access to credits. Credits are awarded to researchers and research projects according to their requirements. Each user has 500 credits available in Transkribus at the beginning. Research projects up to PhD are funded by READCOOP.

Transkribus can also be used in teaching and has proved very useful for decentralised group work on documents.

Topic Modeling

We are working with various research projects and are in the process of setting up a service for Topic Modeling. Topic Modeling: Topics are recognised on the basis of statistical models in large amounts of text. If you are interested in analysing and possibly visualising large amounts of text data on the basis of LDA (Latent Dirichlet Allocation), LSI (Latent Semantic Indexing) or HDP (Hierarchical Dirichlet Process), please contact:



Screenshot: Beispiel Topic Modeling

Screenshot: Beispiel Topic Modeling