MADs Reconciliation Project

File Listing

MADs-reconcile

• app.py
  • Flask application home.
• mads_parse.py
  • Parses XML file for faculty names and URIs and stores in an internal list. Contains functions for searching for a URI.
• requirements.txt
  • Python module requirements

source_files

• ETD-NAF_mads_20220222.xml
  • Contains committee member names and URIs
• pdfdata_names_2021Su.dsv
  • Pipe delimited file containing committee member names, roles, and associated thesis/dissertation PDFs. Was used for testing purposes

Requirements

• Application module requirements are specified in requirements.txt.
• Run the following command to install all required packages: pythom -m pip install -r requirements.txt
• An additional requirement is that the XML file to be parsed for names and URIs must be in the same format as the ETD-NAF... file, as the parser is looking for specific tags in the file.

Usage

• To run reconciliation Flask service run python app.py while in MADs-reconcile directory. If passing in XML document as command line argument, run with python app.py <xml file>. Make sure XML document is in the same directory as the application. Additionally, do not run with flask run, as it messes up command line arguments.

• With OpenRefine running, the Flask application running, and a project opened, run the following steps:

  1. Select a column containing names in indirect order, ex. Smith, John L.
  2. Click reconcile
  3. Click start reconciling (Proceed to Step 5 if service is already added)
  4. Select "Add Standard Service" and supply the following URL http://127.0.0.1:5000/reconcile/mads.
  5. Select "MADs Reconciliation Service"
  6. Supply max # of candidates to return (default is 3)
  7. Select "Start Reconciling..."

• After the reconciliation service finishes, perfect matches with a score of 1.0 will be automatched to their correponding cell. Matches with a score of .8 and a score of .75 are not exact matches, but pretty close, likely matching on the last name and the first name. Matches with a score of .5 only matched on the last name, so confidence is pretty low that these are the correct matches. Matches with a non-perfect score must be manually reviewed to select the best match.

• If only one match is found for a name, the application will automatch it. Comment this out if desired.

• Additionally, if two matches are found for a name, the application will automatch on the one with a higher score. Again, comment this out if desired.

Notes

• This project is based off of source code from mphilli's LoC Reconcilation project. Please see License.
• Developed for Florida State University's Library Technical Services.