AI as research partner: How machine learning is opening medieval archives to global scholarship

Jeffrey C. Witt (Loyola University Maryland)
https://jeffreycwitt.com | jcwitt@loyola.edu

March 19, 2026, Providence College, Providence, RI

Slide Deck: http://jeffreycwitt.com/slides/2026-03-19-providence

https://creativecommons.org/licenses/by-nc-sa/4.0/

## Outline * Introduction * Part 1: Marking Sources in Scholastic Texts * Part 2: Using Marked Sources for Discovery and Exploration
## Introduction
## Thesis 1 My thesis today is simple: if we continue to treat critical editions as isolated, fixed presentations of texts—books or standalone websites—we will not be able to scale our access to the historical record. To take full advantage of modern computational and AI tools, we need to move away from thinking editions as finished, static presentations to treating them as **shared layers of structured data** that are continuously open to use, reuse, and extension.
## Thesis 2 This task is too big because **we are thinking too small**.
## Field Standards * How do we name files * Where do files live * how do we organize files * How do we structure the contents of files --- These banal and boring questions are crucial. It is this organization that allows us to apply automated tools. If they are not answered, then we miss out on the ability to automate workflows and scale our work. --- The specific answer we give is not that important. What is important is that we have the **same** answer.
## Creating the Apparatus Fontium: ### Marking Sources in Scholastic Texts
### Example of Las Casas Canon Law Reference "ff., De reg. iur, cum inquit"
Steps of source identification: * Parse reference into components (author, work, book, chapter, etc.) * Add components to query * Query database * Filter results for best candidate * Insert Reference

Using the Citation Sediment Layer

1. to construct human readable citations and footnotes instantaneously; creating references that are windows and not just pointers.

2. to create deep links into an edition's generative history, allowing users to transparently see all the manuscript witnesses to a given quotation and its sources.

3. to auto-generate indices of sources AND indices of influences (future use)

4. to use these indices to build recommendations of related passages for discovery and exploration.

## 1. Instantaneous Citations and Footnotes
## Citations as a human readable list of containers Work, Book, Chapter Or 1959 edition, volume 35, p. 123.

The problem with container based references

1. Container references are not sufficiently granular and imprecise.

Text Hierarchy Reference (e.g. Work, Book, Chapter) are not sufficiently granular and do not provide enough information to find any specific witness

Material Hierarchy References (e.g. 1959 edition, volume 35, p. 123) are also not sufficiently granular and their page references do not apply to other witnesses.

2. Resolving container based references is laborious and has to be repeated for every resolution.

## Citations as granular unit level unique identifiers with hierarchical relationships

We used to use containers to find granular places

Now we use the abstract id of granular places, and let the places locate themselves along with all their containers

### A GPS coordinates for our texts
## 2. Deep Links into the Generative History of an Edition
## 3. Auto Generated Indices of Sources and Influences
## 4. Using Indices to Build Recommendations of Related Passages for Discovery and Exploration

Final Thoughts

## Threefold Mission of the SCTA 1. To develop field standards 2. To aggregate data created according to field standards 3. To publish aggregated data for open and creative re-use by the community.
## Thank You ### Questions and Comments
## Scholastic Commentaries and Texts Archive
### SCTA as a new kind of Scholarly Society
## Threefold Mission of the SCTA 1. To develop field standards 2. To aggregate data created according to field standards 3. To publish aggregated data for open and creative re-use by the community.
### The "Platform Effect"
## Thank You ### Questions and Comments?