We develop and run code on a variety of platforms, mainly laptop and desktop computers. Our local number-cruncher is an IBM ThinkCentre equipped with an NVIDIA GPU.
For programming IDE, we recommend Visual Studio Code (vscode) which we run on both Mac OS and Ubuntu.
We run python 3 in virtualenvwrapper environments.
For simpler natural language processing tasks, we rely heavily on spaCy and use corpora and simple tools from NLTK.
For classification, we use SimpleTransformers which builds upon the comprehensive Hugging Face Transformers library.
For parsing LaTeX, we use pylatexenc which is based on TeXSoup.
For parsing XML and HTML, we use Beautiful Soup.
For statistical analyses, we use python modules for initial processing. Subsequent, deeper analysis is either tailored in python or done in R.
For qualitative analyses we recommend NVivo12 (UCPH licence).
To support project management, we use MindManager (UCPH licence), Trello and dh4pmp.slack.com.
For image detection, we use the Detectron2 framework.
For handling PDF files, we rely on fitz aka PyMuPDF as well as PyPDF2.
For image tagging, we use a customized version of LabelImg, which supports multiple attributes.
Most of our developed software is hosted in a private SubVersion system.
However, some of it is copied to a private GitHub for ease of access within the group, including:
If you want access to any of these repositories, get in touch.And we also contribute to other projects, including: