ma-thesis-dh-code

Repository Description

This repository provides code and material that is used in my master's thesis ("Quantifying Formulaic Flexibility of Middle High German Legal Texts"). It is programmatically functional and primarily serves the exploration of data. As such, it is to be read in conjunction with the study, and vice versa. Find the text on https://doi.org/10.5281/zenodo.8141830 or https://unipub.uni-graz.at/obvugrhs/8653213.

File Description

py/

• py/.ipynb contains notebooks used for the analysis and visualization of data. Except 0x, they are numbered according to the research questions
• py/ddp_util.py and formutis.py contains helper functions

Setup

Main

• Clone repo, create (interactive) Python environment (>=3.8; 3.10 recommended) under Linux or WSL
• Install additional dependencies depending on notebook (full requirements.txt is likely overkill).

Optional

For transforming reference corpus

• Fully unzip REM tarball in /corpora/REM/ (or download via https://zenodo.org/record/3624693)

For custom language detection of historical German

• Download https://huggingface.co/ERCDiDip/langdetect into models/mom-langdetect

For sequence prediction

• Download https://huggingface.co/atzenhofer/distilroberta-base-mhg-charter-mlm into models/custom/distilroberta-base-mhg-charter-mlm
• Download https://huggingface.co/atzenhofer/xlm-roberta-base-mhg-charter-mlm into models/custom/xlm-roberta-base-mhg-charter-mlm

Misc

Formatted using mostly https://github.com/psf/black and https://github.com/csurfer/blackcellmagic