This repository contains code related to the "Encoding Reparative Description" project, which began in 2023 with a team of researchers at the University of Michigan School of Information, the Bentley Historical Library, and the University of Michigan's Humanities Collaboratory.
There are two basic inputs for this tool, a required term list and an optional ArchivesSpace Resource ID list.
The required term list be a .txt file with one term per line, e.g.:
Civilization
Civilized
Uncivilized
Burial
Burials
...
See terms-nativeAmerican.txt and terms-phillipines.txt as examples.
Optionally, you may also provide a second .txt file with one ArchivesSpace Resouce ID per line, e.g.:
229
8482
4460
256
8480
...
If the user does not provide a .txt file with a list of Resource IDs, the tool will parse ALL ArchivesSpace resources. Since this is so resource intensive, the tool defaults to expecting a .txt file.
Run the parse_resources.ipynb Python notebook first. It takes the inputs above, along with a configuration file described in the notebook, and uses them to parse ArchivesSpace via the API. It produces a .csv file used in later operations--see results-nativeAmerican.csv and results.phillipines.csv as examples.
Note that kwic_report.ipynb has been deprecated. Run the key_term_in_context.ipynb Python notebook to create a key term in context report from the results. See matched_results-nativeAmerican.csv and matched_results-phillipines.csv as examples.
Note that visualize_matches.ipynb has been deprecated. I think. Run the match_visualize.ipynb Python notebook (or its companion, match_visualize.html) to create visualizations from the results. Optionally, save individual visualizations as .png files. See the following examples:
element_frequency_accross_dfselement_frequency_accross_dfs_stackedterm_frequency_by_repoterm_frequency_by_repo_horizbar