Matteo Di Cristofaro

I love this!

Screenshot of the app showing a page from a book + different views of existing and new ocr.

Many VLM-based OCR models have been released recently. Are they useful for libraries and archives?

I made a quick Space to compare VLM OCR with "traditional" OCR using 11k Scottish exam papers from @natlibscot.bsky.social

huggingface.co/spaces/davanstrien/ocr-time-capsule

rvya redeemed, thanks

b83m redeemed, thanks a lot

fsdc redeemed, thanks!

4-panel comic. (1) [Person 1 with ponytail flanked by person with short hair and another person speaking into microphone at podium] PERSON 1: In the early 2010s, researchers found that many major scientific results couldn’t be reproduced. (2) PERSON 1: Over a decade into the replication crisis, we wanted to see if today’s studies have become more robust. (3) PERSON 1: Unfortunately, our replication analysis has found exactly the same problems that those 2010s researchers did. (4) [newspaper with image of speakers from previous panels] Headline: Replication Crisis Solved

Replication Crisis

xkcd.com/3117/

How social media destroys democratic discourse, explained in 6 easy figures Where we all went wrong

I believe it is worth interrogating the fundamental forces re-shaping our information spheres away from liberal democracy towards myth, manipulation and magical thinking empowering autocracy and nihilism.

Here’s how it all falls apart—a 🧵 in 6 figures ⬇️
www.protagonist-science.com/p/how-social...

Regulating AI Isn’t Enough. Let’s Dismantle the Logic That Put It in Schools. AI in schools isn’t progress — it’s a sign of how far we’ve strayed from the purpose of education.

Stuffing ai into everything “isn’t just a forecast, it’s a libidinal fantasy — a capitalist dream of replacing relationships with code and scalable software, while public institutions are gutted in the name of ‘innovation.’”

Companies That Tried to Save Money With AI Are Now Spending a Fortune Hiring People to Fix Its Mistakes Companies that rushed to replace human labor with AI are now shelling out to have IRL workers to fix the technology's screwups.

🤷🏿‍♂️

"The problem with AI isn't that it can do your job. It can't. The problem with AI is that your MBA-brained boss's boss doesn't know how your job works and thinks AI can do your job at fractions of a penny on the dollar, and hears the siren song of 'maximize shareholder value'."

MBA-brain is real.

Data interference: emojis, homoglyphs, and issues of data fidelity in corpora and their results Tokenisation - "the process of splitting text into atomic parts" (Brezina & Timperley, 2017: 1) - is a crucial step for corpus linguistics, as it provides the basis for any applicable quantitative met...

Is 😵‍💫 one token or two?
To a human, it's one. To a corpus tool, it’s often split (😵 + 💫).
And 𝙊𝙉𝙇𝙄𝙉𝙀 ≠ online.
This preprint shows how emojis & homoglyphs challenge tokenisation and distort linguistic evidence.
🔍 arxiv.org/abs/2507.01764

#Emoji #Homoglyphs #CorpusLinguistics #AcademicSky #LangSky

wow, many thanks!

Fellow academics, can anyone help with obtaining an #endorsement on arXiv?
I have a preprint I'd like to upload to Computer Science > Computation and Language (cs.CL), but need someone to endorse my account.
Here's the endorsement link: arxiv.org/auth/endorse...

#corpuslinguistics #linguistics

3jkl redeemed, thanks

y4ha claimed, thanks!

Memes can serve as strong indicators of coming mass violence A new study finds that surges in visual propaganda—like memes and doctored images—often precede political violence. By combining AI with expert analysis, researchers tracked manipulated content leading up to Russia’s invasion of Ukraine, revealing early warning signs of instability.

Memes can serve as strong indicators of coming mass violence

Finally, a Replacement for BERT: Introducing ModernBERT We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Finally, a Replacement for BERT (Blog about ModernBert)

huggingface.co/blog/modernb...

📒The scientific program is out!

Click on the link below to have a look at the speakers and the workshops of our Summer School!

⬇️⬇️⬇️

www.summerschooldigitalhumanities.unimore.it/2025-edition...

Our Summer School is beginning now with the institutional greetings.

Happy Graduation to all students and staff!

Postdoc position open in Zurich -- Prof. Martin Tomasik and I have a joint SNF project on interpretable neural network approaches for large scale, complex item / temporal structure, online learning / cognitive development data.

Please retweet.

tinyurl.com/PostdocGNNSNF

xsds redeemed.
now I just need to finish packing and I'm ready for the woods!

fy8n redeemed, perfect scent!

3eyg redeemed.
the healing begins

P hacking — Five ways it could happen to you Some data practices can lead to statistically dubious findings. Here’s how to avoid them.

⏹️ Ending the experiment too early
🎯 Running experiments until you get a hit
🍒 Cherry-picking your results
🔧 Tweaking your data
➗ Not adjusting for multiple comparisons

www.nature.com/articles/d41...

Academics and Universities have got to formulate a coherent approach to AI and guide their students. Cries of despair and digitally illiterate pronouncements will not reverse the effects of technical innovations.

YouTube video by Edinburgh ReproducibiliTea Reproducibilitea – Designing a Good Research Practice Roadmap

Designing a Good Research Practice Roadmap
- a presentation by Fiona Ramage @fionar.bsky.social
youtu.be/dnzRQPOxz1o?...

The event was organised by Edinburgh ReproducibiliTea

With Canada’s election just days away, the continued appearance of deepfake ads reveals a serious flaw in Meta’s ad-review system. If fraudsters can repeatedly bypass detection, it suggests the platform’s current safeguards are not equipped to catch even basic forms of manipulation...

4chan Is Dead. Its Toxic Legacy Is Everywhere It’s likely that there will never be a site like 4chan again. But everything now—from X and YouTube to global politics—seems to carry its toxic legacy.

"Twitter became 4chan, then the 4chanified Twitter became the United States government. Its usefulness as an ammo dump in the culture war was diminished when they were saying things you would now hear every day on Twitter," @bencollins.bsky.social told WIRED.

Delighted to share my newly published review of "Data Analytics for Discourse Analysis with Python" (Tay 2024). The work addresses urgent, discipline‑wide concerns in #linguistics, and writing about it was both a privilege and a joy.

authors.elsevier.com/a/1kyHu1L-nh...