Project Overview
This digital collection emerged from doctoral research examining the intersection of digital humanities methods, archival science, and diplomatic history. The project addresses a fundamental challenge in historical scholarship: how to make large-scale archival collections computationally accessible while preserving the integrity and context of primary source materials.
Working with 13,848 pages from Library and Archives Canada's microfilm holdings, this research develops and tests automated metadata extraction pipelines designed specifically for historical diplomatic correspondence. The resulting dataset enables new forms of computational analysis—including network mapping of diplomatic communications, temporal pattern detection, and quantitative approaches to wartime diplomacy—while maintaining the scholarly rigor required for historical research.
Methods and Innovation
The project employs artificial intelligence-powered Optical Character Recognition (OCR) optimized for typewritten historical documents, achieving significantly higher accuracy rates than traditional OCR methods on mid-20th century materials. Through iterative development of automated metadata extraction algorithms, the system aims at identifying and structuring key information including sender/recipient locations, message dates, file references, and security classifications with minimal manual intervention.
This computational approach represents a contribution to digital humanities: demonstrating how carefully designed automation can scale archival processing while producing research-grade structured data. The project's code, documentation, and quality assessment frameworks are being developed as reusable tools for similar historical document collections.
Collaboration and Academic Context
This work is conducted at the University of Ottawa by Vincent Martin-Schreiber in collaboration with Florian Mathieu (Université Paris-Saclay).
The research contributes to ongoing scholarly conversations about Vichy diplomacy, Canadian signals intelligence history, and the application of computational methods to archival research. Results and methodologies are being prepared for presentation at international digital humanities conferences and publication in peer-reviewed journals.
Open Access and Future Directions
All materials are released under a Creative Commons Attribution 4.0 International License, supporting the principles of open scholarship and enabling reuse by researchers worldwide. Future phases of development will incorporate Named Entity Recognition for diplomatic personnel and locations, enhanced network analysis tools, and collaborative features allowing researchers to contribute corrections and annotations.
This project demonstrates the potential for computational approaches to unlock historically significant but previously under-utilized archival collections, creating new possibilities for research while respecting the evidential value and historical context of primary sources.
Contact
Project Lead: Vincent Martin-Schreiber
Institution: University of Ottawa
Academic Advisor: Florian Mathieu, Université Paris-Saclay
Email: examination-unit.hardly744@passinbox.com
For technical questions, corrections, or collaboration inquiries, please use the contact information above.