Employing ML Methods on Digitized FOIA Requests for Improved Discoverability and Policy Research

Seaton, Alexa; Xu, Yujie; Von Arx, Devin; Traylor, Jordan; Jin, Ying; Evans, Kenneth Mellinger

Employing ML Methods on Digitized FOIA Requests for Improved Discoverability and Policy Research

Files

seaton-TXLA.pdf (1.36 MB)

Date

2025

Authors

Evans, Kenneth Mellinger

Publisher

Rice University

Abstract

Born-digital records pose challenges for digital preservation due to their unstructured formats and noncompliance with accessibility standards. This project introduces a modular, open-source workflow to batch process large, mixed media PDFs—many obtained through FOIA requests—by leveraging OCR, AI, and named-entity recognition. Built for the White House Scientists Archive, this system enhances discoverability and usability of digitized records across administrations and supports metadata extraction at scale. Key tools include Mistral AI for OCR, Apache Tika for entity recognition, and a finet uned Mistral model for metadata generation.

Type

Presentation

Citation

Seaton, A., Xu, Y., Von Arx, D., Traylor, J., Jin, Y., & Evans, K. (2025). Employing ML Methods on Digitized FOIA Requests for Improved Discoverability and Policy Research. Rice University. https://doi.org/10.25611/AYBM-0F31

Published Version

https://doi.org/10.25611/AYBM-0F31

Rights

Except where otherwise noted, this work is licensed under a Creative Commons Attribution-NonCommercial (CC BY-NC) license. Permission to reuse, publish, or reproduce the work beyond the terms of the license or beyond the bounds of fair use or other exemptions to copyright law must be obtained from the copyright holder.