The University of Michigan (U-M) Library’s Library Information Technology (LIT) Division has collaborated with the Institutional Data Initiative (IDI) at the Harvard Law School Library to transform the way we view university history.
Since 1837, the U-M Board of Regents has documented the university’s important governance-related decisions in their meeting minutes. The U-M Library began hosting digital versions of the Board’s meeting minutes over two decades ago, creating the library’s Proceedings of the University of Michigan Board of Regents collection, whose content is dedicated to the public domain under Creative Commons CC0 License by the Regents of the University of Michigan.
Thanks to the pilot project that started in 2025, these historic records are now being re-released as datasets with advanced AI enhancements, which will make it much easier for scholars and researchers to query and find information within them.
What has changed?
- Better Organization: Over 55,000 scanned pages from 71 volumes have been organized into 1,755 distinct individual meetings with re-OCRed text and page layout information.
- Richer Descriptive Metadata: Every meeting record is now tagged with details that have been historically hard to discover but frequently sought out by researchers and university staff, such as specific meeting dates, individual attendees, and key topics that include changes in the university’s leadership or degree programs, new schools and colleges, building projects, funding campaigns, and strategic plans.
- AI Summaries and Syntheses: Beyond the original text and enriched metadata, the datasets provide summaries per meeting and for each decade as high-level overviews of how the university has evolved over time.
- Accessibility Improvement: The records for individual meetings are now also available in the EPUB format following the latest digital accessibility standard, making them more accessible to human readers including those with visual impairments.
Where to find the datasets?
- The new datasets are available at the U-M Deep Blue Data Repository (forthcoming), ICPSR (forthcoming), and Hugging Face.
- The technical report describing project work in detail is available at the IDI website (forthcoming on arXiv), and the processing pipeline used for this project is also shared openly on GitHub.
Next steps
- The U-M Library’s Library Information Technology (LIT) Division will be leveraging these AI-enhanced datasets to make improvements in the accessibility, browsability, and discoverability of the Proceedings of the University of Michigan Board of Regents digital collection in the new digital repository platform.
- The new digital repository platform is to be completed by 2027 through the project work that began in 2025 under the name of Project DOR (Digital Object Repository). The new repository will be equipped with functionalities to better serve and preserve all types of digital content to meet the evolving needs of scholars and researchers worldwide.
The work on the Proceedings also points to possibilities for developing AI methods that can more accurately and reliably surface and extract the traces of human decision-making in the historical record, which will shed more light on how human institutions, and history itself, have unfolded and evolved over time.
We hope this pilot project provides libraries and archives with an opportunity to think more deeply and creatively about ways in which a wide variety of content under their long-term care and stewardship may be enhanced and made into more accessible and readily usable source materials for both traditional and computational research methods. You can also view the IDI’s announcement on this joint pilot project on their website.
Questions?
For questions, please contact Bohyun Kim, Associate University Librarian for Library Information Technology or Chris Powell, Coordinator of Encoded Text Services, in the Library Information Technology division of the U-M Library.