#chr2024 — Public Fediverse posts

My own talk, which is coming up / just happened, is on #Multilingual #Stylometry.

Based on our work for #CHR2024, we've moved on from the influence of language and translation on stylometric attribution accuracy to the influence of corpus composition. I'll be presenting both parts, but for the corpus composition issue, we are still at the stage of preliminary results.

With a special thanks to @artjomshl.bsky.social!

Slides: https://dhtrier.quarto.pub/icla/

@rebsim #ICLA2025

@jomla I agree that at #DHd2025, #Mastodon worked great. Similarly at #CHR2024 last year. It is more quite here at #DH2025, clearly. But I hear that #Bluesky is not all that intense either, apparently? (Maybe only in the first few days.) I think #Whova takes some of the social media / microblogging energy away from the open social media, which is a pity.

A point of consensus that emerged tacitly in #CHR2024: memorization is not creating big distortions for literary research using LLMs. Eg in paper #96, Zhang, Seminck, & Amsili conclude "overall, the degree of memorization was low." And echoed later in same panel by @dbamman.bsky.social. +

ceur-ws.org/Vol-3834/paper...

Awesome! The #CHR2025 edition will take place at #C2DH in #Luxembourg! That's a great place for this conference (and right around the corner from @tcdh at #Trier University).

"Small country, big ambitions."

#CHR2024

#CHR2024 is almost over... it was an awesome conference! Very high quality of the talks, great food, wonderful people!

(The only thing I have not been impressed with is that #CHR2024 is definitely, and sadly, NOT happening here on #Mastodon.)

Heard about so many models, tools and methods here at #CHR2024: Could I please have a few months off to try them all out on our various datasets?

Here's the link to the abstract and paper: https://2024.computational-humanities-research.org/papers/paper102/

And here's the link to an interactive showcase, where you can play around with the data: https://edabel.shinyapps.io/setlist-variety/

Of course, the data is there as well: https://github.com/edabel/setlist-variety

#CHR2024

💯 I honestly did not know what to expect from my last talk at #CHR2024, on some statistics regarding setlists of bands on their tours, but blown away by the actually really fascinating trends and patterns in the data: Which songs are played almost always, which ones only once, when looking at the shows in a tour. Different shapes for different bands at different times, and interesting patterns where cover songs are used: in the shelf (core repertoire) or the tail (singletons). Kudos to Ed Abel!

Now up at #CHR2024, a paper on historical "arrest patterns" in Brussels, by Folgert Karsdorp @folgertk, Mike Kestemont and Margo de Koster.

Abstract and link to full paper here: https://2024.computational-humanities-research.org/papers/paper13/

(Image is not about arrests, but first slide to illustrate diversity in a collection, which is their approach to the data.)

The #multilingualism sesssion at #CHR2024 is going strong!

Fascinating talk by @laurenfklein at #CHR2024! She explored the influence of (Critical) #Theory on modeling in the #DigitalHumanities, emphasizing theory-driven modeling and the interpretation of such models as fundamentally humanistic endeavors. Highlights included her work on social media discourses around #BlackLivesMatter, abolitionist networks in historical newspapers, and more. #DHTheorie #DHTheory

It’s alive! In our project „Animacy in German Folktales” we annotated animacy in a literary corpus and built a classifier. Results where just presented at the wonderful #CHR2024 in Aarhus, Denmark. Keep on reading for a brief summary of the key arguments. #DH #CLS @comphumresearch (1/6)

Congrats, @josecalvo, for this review of your book!

"It is the reference works of this kind, with a rigorous methodological approach, that predict a prolific and promising future for the development of digital humanities."

In the book, José uses corpus and computational methods for genre analysis in the Spanish novel of the Silver Age:

Full review by Isabel Moyano Moreno (in Spanish): https://revistascientificas.us.es/index.php/PH/article/view/25147

José's book (OA): https://www.bielefeld-university-press.org/978-3-8376-5925-2/the-novel-in-the-spanish-silver-age/?number=978-3-8394-5925-6

@fotis_jannidis @jcls #CHR2024

Now up, at #chr2024 "Explainable Search and Discovery of Visual Cultural Heritage Collections with Multimodal Large Language Models" by Taylor Arnold and Lauren Tilton!

About the "#Documerica" collection of photos from the 1970s. Fascinating material!

Check out the collection here: https://www.archives.gov/research/environment/documerica-topics

Important question by David Bamann at #CHR2024: "When do we need LLMs? When do simpler, interpretable models suffice?"

Mentions computational and environmental cost as things to consider.

Really interesting work on fanfic vs. popular novels and writing style vs. ratings 2024.computational-humanities-research.org/papers/paper... #fandom #chr2024

Cool talk at #CHR2024 by Daniil @skorinkin of the #Potsdam team, on the literary geography of Russian nineteenth-century novels. Some cities lose importance over time, others gain importance, and the overall trend appears to be an "outward turn" (see slide).

@jacomyma Agreed!

I mean, there are multiple things we can do with LLMs, but their strength is generating text, not numbers. It's just that free prose as a result has its limitations as well.

To be fair, the radio archive use case did just that, adding descriptive phrases and keywords to audio files.

I actually think that even for free prose text, using LLMs to generate explicit, potentially semantically modeled information, is an important application. Then, we can count stuff...

#CHR2024

Very nice lightning talk by Oleg Sobchuk et al. on literary evolution, at #CHR2024!

Large HathiTrust corpus and pretty fancy implementation modeling attention decay and potential intertextual links between texts in a graph-based model, then make this dynamic.

Later today at #CHR2024, we are going to present our work on #Multilingual #Stylometry!

We isolated the influence of #language on #authorship #attribution #accuracy by translating multiple #corpora into each others' languages while keeping #corpus composition stable.

Interactive showcase: https://showcases.clsinfra.io/stylometry

Full paper: https://ceur-ws.org/Vol-3834/paper9.pdf

This work was developed within the @CLSinfra project in #Trier, #Krakow and #Prague with Artjoms Šeļa, Evgeniia Fileva and Julia Dudar.

The second day of #CHR2024 has started. The first session is one with 18 5-minute lightning talks! That's going to be a roller coaster! 🎢

💡 A rich and wonderful program at the Computational Humanities Research #CHR2024! Kudos to the organizers @comphumresearch, #AarhusUniversity! Many facets of computational approaches in the Humanities are being discussed. Looking forward to the Lightning Talk session shortly and excited to contribute myself: "Second-Order Observation through AI: Towards a Humanistic Approach to Augmenting Human Intellect."
👉 http://2024.computational-humanities-research.org/programme/#lightningtalks
#Hypertext #AI #ComputationalHumanities #DigitalHumanities

Such a great setting for a dh conference reception, museum of antiquities in Aarhus. Plus colleagues from afar. ❤️ #chr2024