Search

The use of High Performance Computing (HPC) offers a wide range of possibilities in the digital humanities. The following sections list possible applications.

We are looking forward to supporting you with your research project. If you have an idea for an application or a project or simply want to try something out, please contact us. We will help you get started and implement it with our HPC systems.

Together with the research development team at the SUB Göttingen, we offer customised software solutions for scientists and their research projects. To this end, we advise on the selection of suitable technologies and jointly develop digital tools for natural language processing, language models and artificial intelligence as well as for the modelling, processing and visualisation of research data.

Text Processing

HPC can support text processing, from simple data transformations to the creation of complex models.

Possible applications:

  • XML transformation of many files
  • Supplementing data from open data sources
  • Automatic classification of texts
  • Finding similarities in writing styles
  • Automatic classification of persons, places, and time periods
  • Generation of automated summaries of texts

Automated Capture of Metadata

HPC can support the automated capture of metadata for image, video, and audio media.

Image and Video

Possible applications:

  • Recognition of scenes (landscape, portrait, building, …)
  • Recognition of shapes
  • Person and face recognition
  • Color spectrum
  • Identification of variants and manipulations

Audio

Possible applications:

  • Transcription of speech into text
  • Characterization of voice and mood

Digital Restoration and Computer Simulations

HPC can support the use of digital technologies for the restoration and simulation of objects, scenarios or content.

Possible applications:

  • digital reconstruction of cultural assets
  • 3D scans of objects
  • Simulation of climate changes
  • Simulation of storage conditions for cultural assets such as paintings

Ontologies and Knowledge Graphs

HPC can support the construction and analysis of large knowledge graphs. Designing the ontology itself needs no cluster — applying it at scale usually does.

Possible applications:

  • Extraction of entities and relations from large corpora into an ontology (e.g. CIDOC-CRM)
  • Linking entities to authority files (GND, Wikidata, VIAF, GeoNames)
  • Alignment of vocabularies and ontologies
  • Reasoning over large graphs (memory-bound rather than parallel)
  • Validation of graphs with millions of triples
  • Link prediction, e.g. suggesting missing attributions or provenance connections
  • Network analysis of correspondence, citation, or provenance graphs

If you can do it in a local triple store, you probably don’t need HPC — ask us if unsure.

Services provided at GWDG

  • Jupyter HPC with GPU support (A100/H100)

We provide access to Jupyter-based HPC environments with GPU acceleration. This setup is ideal for tasks such as text mining, data visualization, and large-scale digital archives. For example, researchers can use GPUs to speed up processing of historical texts, analyze linguistic patterns in vast corpora, or visualize complex data from digital heritage projects.

  • Chat AI with Large Language Models (LLM))

Our platform supports the use of LLMs for applications like content generation, sentiment analysis, or language processing. You can easily access our API via Jupyter, enabling you to integrate powerful AI tools into your digital humanities workflows, whether for analyzing historical documents or creating interactive AI-based content.

  • Fine-tuning or training language models for tasks such as neural machine translation of indigenous or low-resource languages using parallel corpora
  • Processing large document collections at scale for OCR, HTR (Handwritten Text Recognition), image extraction, layout detection for analysis of historical newspapers and archival records. (E.g HTR on 40,000 scanned pages: 6 hours on HPC with one A100, versus two weeks on local computer."
  • Stylometric authorship attribution applied simultaneously across thousands of texts
  • Training or fine-tuning computer vision models on art history datasets (paintings, iconography, manuscript illustrations), or running inference through existing vision models on the cluster via the SAIA API

Talk to us

Please contact us if you have an application idea, project idea or just want to try something out. We will assist you to implement this, even if you have no previous experience on the HPC.

If you would like to get started on your own, our Academy offers introductory courses on HPC and GPU computing.