As part of the BeCoRe project, we tried to create a script that would automatically identify the issuers of acts on the basis of their regeste.
Runs with python 3.10 and packages listed in requirements.txt
This script is based on the spacy package and requires the installation of the language models that will be used: https://spacy.io/usage/models/.
Please note: we have only used the 'lg' versions of the language models proposed by spacy.
This script was used with two corpora from monasterium.net: 1 - the acts of the abbey of Fontenay (encoded by Dominique Stutzmann) with regests in French 2 - the acts of the abbey of Gottweig (whose issuers are encoded by Jacqueline Schindler) with regests in German.
Several trials were carried out and their results are stored in eponymous files. The current .csv files are the results of the last test with the Gottweig acts.
The script is able to detect the language of the source file, but the current version only runs on French and German files.
You can use this script on command line with regests contained in a cei file:
python main.py "cei-file.xml" The result is contained in 2 files: 1 - "results.csv" containing the regests with the issuer in a dedicated column and a comparison with the issuer already encoded if applicable 2 - "unused.csv" containing the regests that have been detected as unusable for the script.