Thibault Clérice

Poster advertising lectures on "Raisonnement Philologique et Modèles Informatiques" stating at 4pm, Thursday, March 12, at 54 Boulevard Raspail, Paris.

Paris friends! Amis parisiens ! This Thursday is the first of four public lectures I'm giving on AI and philology, broadly defined: "Philological Reasoning and Computational Models." The advertisement is in French, but the lectures are in English. I'd also love to meet while I'm here in March! 1/

@transkribus.bsky.social Is there a public API to access our documents and documents export, to streamline their publication ? I am not able to find any information on such an option that is not > 2 or 3 years.

Excellent venue for computational humanities work, colocated with ACL in San Diego on July 6. Please share!

📣 Dans le cadre de l'EquipEx+ Biblissima+, l' @ensdelyon.bsky.social recrute un.e ingénieur.e d'études chargé.e de l'édition de corpus numériques.

🗓️ Postulez à cette offre avant le 1er avril 2026 👇️
ens-lyon.softy.pro/offers/196330

@irht-cnrs.bsky.social
@mrshcaen.bsky.social

We’re looking for a Research Data Engineer (m/f/x) (3,5 years). If you have a #DigitalHumanities profile with experience in #TEI encoding, #OCR / #HTR, and #IIIF (or any of those and are willing to learn the rest), get in touch! The full time position can be split, so if (for whatever reason) […]

Workshop (Call for proposals) – Analysing Cultural Heritage Documents: HTR/OCR, Information Extraction, and Textual Variation Workshop (Call for proposals) – Analysing Cultural Heritage Documents: HTR/OCR, Information Extraction, and Textual Variation  AI for Cultural Heritage Documents: HTR/OCR, Information Extraction, and ...

Workshop (Call for proposals) – Analysing Cultural Heritage Documents: HTR/OCR, Information Extraction, and Textual Variation prima.hypotheses.org/3737

We have this experience as well, a paper is coming up later this week with details about a specific domain...

Hold my non alcoholic beverage 😁
It's not on my to do list but if a team wants to look at it with us...

Hotel de la région, région qui penche bien a droite depuis des années, pas étonnant non ?
Dégoûtant, mais pas étonnant

Y a une page wikipédia extrême droite à Lyon. On en est là fr.wikipedia.org/wiki/Extr%C3...

La région Rhône Alpes quoi...

Setting aside the original issue, on which we agree 😉, you are right. This particular manuscript and another have no open data in this paper apparently...

Or, an error in the paper presentation (not unseen in the humanities ) where they make us expect consecutive line by design and this is not. I would rather expect this than bad curation given that they actually spend time looking at errors.

Thank you for the catch ! Should have been more careful here...

Manuscrits de la Médiathèque du Grand Troyes. Manuscrits issus de la bibliothèque de Clairvaux. Ms. 1600 Manuscrits de la Médiathèque du Grand Troyes. Manuscrits issus de la bibliothèque de Clairvaux. Ms. 1600 -- -- manuscrits

You are completely right. I was so baffled by the dya9 bad interpretation that I just did not check the rest...
The original paper by Aguilar showed stitched lines most of the time and trusted it...
Paper(P.17): hal.science/hal-04716654/
Manuscript: gallica.bnf.fr/ark:/12148/b...

Digital Classicist London 2026 Call for Papers #cfp

blog.stoa.org/archives/4370

And for those who want the answer:

❌ dyaconus -> diabolus
❌ in futurum -> infusum

There are plans :)

We also show that we are far from done, specifically for a complicated language like Old French.

But we
(1) defined the issue,
(2) propose a first solution that enables pre-annotation of larger dataset and
(3) offer an alternative to less trustable models that go beyond ATR.

Pre Editorial Normalization - a Hugging Face Space by comma-project Latin and Old French normalization of CATMuS output

We release:

📚 4.66M silver training samples
🧪 1.8k gold evaluation set huggingface.co/datasets/com...
🤖 ByT5-based model → 6.7% CER huggingface.co/comma-projec...

Try it here 👇
huggingface.co/spaces/comma...

👉 We propose Pre-Editorial Normalization (PEN):

An intermediate layer between:
📝 graphemic ATR output
📖 fully edited text

Goal: preserve palaeographic fidelity + enable usability.
Keep two layer, ATR output and normalization, with aligned token to go back to the source.

Recent ATR progress—especially with palaeographic datasets like CATMuS—has improved access to medieval sources.

But:
❌ Raw outputs are hard to use
❌ Fully normalized models over-normalize & hallucinate

There’s a methodological gap.

If I give you the text
📚 omnium peccatorum quia ex quo dyaconus quando esset in futurum, stultus esset

Can you find the ATR error without the manuscript ?

Probably not.

ATR models that predict text and normalize in one go generate trustable text, but prevent detecting issues.

📄 New paper:
Pre-Editorial Normalization for Automatically Transcribed Medieval Manuscripts in Old French and Latin

Thibault Clérice, @rachelbawden.bsky.social , Anthony Glaise, Ariane Pinche, @dasmiq.bsky.social (2026) arxiv.org/abs/2602.13905

We introduce Pre-Editorial Normalization (PEN).

🧵⬇️

Pre Editorial Normalization - a Hugging Face Space by comma-project Latin and Old French normalization of CATMuS output

I got something up, thank you :) huggingface.co/spaces/comma...

Participation in the community call for DTS Please provide your email address below to be receive a link for the community call. All other fields are optional, and just help us know a little more about the community.

To mark this milestone, we are organizing a community call on April 23rd, 15:00 CET / 9:00 EST.
Register here: docs.google.com/forms/d/e/1F...

We are pleased to announce the official release of Distributed Text Services (DTS) v1.0 — a stable specification, ready for broad adoption.

This release is the result of years of collaborative development, and community feedback.

The specification are available at: dtsapi.org/specificatio...

Thanks

Oh ! Sorry, I meant @danielvanstrien.bsky.social and did not see I clicked on the wrong username...