iiinova — where your digital self keeps evolving

DH Group at FBK @dh-fbk.bsky.social · 08/07/2026

We release MRNI-BB to support further work on male-directed gender bias and on how gender-role norms shape LLM inference. Paper + dataset: github.com/dhfbk/MRNI-B...

github.com

GitHub - dhfbk/MRNI-BiasBenchmark

Contribute to dhfbk/MRNI-BiasBenchmark development by creating an account on GitHub.

000

DH Group at FBK @dh-fbk.bsky.social · 08/07/2026

Across models, we find a dissociation: they rarely endorse traditional masculinity norms when asked directly, but often attribute MRNI-aligned behaviors to men in scenario-based inference, especially under ambiguity.

100

DH Group at FBK @dh-fbk.bsky.social · 08/07/2026

We test five open-weight LLMs: Aya-Expanse-8B, Llama-3.1-8B-Instruct, Llama-3.1-70B-Instruct, Mistral-7B-Instruct-v0.3, and Qwen2.5-32B.

100

DH Group at FBK @dh-fbk.bsky.social · 08/07/2026

Building on this framework, we develop MRNI-BB, a new English–Italian scenario-based benchmark. The dataset varies both gender marking and the amount of evidence available, allowing us to test whether models rely on masculinity norms when the context is clear, ambiguous, or uninformative.

100

DH Group at FBK @dh-fbk.bsky.social · 08/07/2026

We ground the evaluation in the Male Role Norms Inventory, a psychological framework used to measure endorsement of traditional masculinity norms, such as toughness, dominance, self-reliance, emotional restraint, and avoidance of femininity. #NLProc #ACL2026

100

DH Group at FBK @dh-fbk.bsky.social · 08/07/2026

Can LLMs encode biased expectations about men? In our ACL Findings 2026 paper, “Real Men are Tough”, we study whether LLMs rely on traditional masculinity norms as expectations about male agents. @satonelli.bsky.social @elisaleonardelli.bsky.social @casulalm.bsky.social @boglarkanyul.bsky.social

102

DH Group at FBK @dh-fbk.bsky.social · 02/02/2026

Consider submitting your work to the Workshop on Perspectivist Approaches to NLP, co-located with LREC 2026!

031

Reposted by DH Group at FBK

land-fbk.bsky.social @land-fbk.bsky.social · 21/01/2026

🇸🇬 Don’t miss @penzo-nicolo.bsky.social presenting his poster on synthetic Multi-Party Conversation generation at #AAAI26! 🏛️ Hall 2 🪧 Poster Board 257 ⏰ Jan. 22, 12–2 PM

012

DH Group at FBK @dh-fbk.bsky.social · 21/01/2026

🇸🇬 @penzo-nicolo.bsky.social will be presenting his poster on synthetic generation of Multi-Party Conversations at #AAAI26, don't miss it! 🏛 Hall 2 · 🪧 Poster Board 257 · ⏰ 12–2 PM, Jan 22nd

000

Reposted by DH Group at FBK

Nicolò Penzo @penzo-nicolo.bsky.social · 13/01/2026

Interested? Come by our Poster @ Poster Session (NLP) 1 on January 22nd, 12–2 PM, Hall 2! See you in Singapore! 🇸🇬 🧵9/9 @dh-fbk.bsky.social @land-fbk.bsky.social @mobs-fbk.bsky.social

021

Reposted by DH Group at FBK

Nicolò Penzo @penzo-nicolo.bsky.social · 13/01/2026

🎶One more “song”? Come see our poster next week at #AAAI2026 on our paper: 📣 “Don’t Stop the Multi-Party! On Generating Synthetic Written Multi-Party Conversations with Constraints” With @marcoguerini.bsky.social Bruno Lepri @gglavas.bsky.social @satonelli.bsky.social 🧵 arxiv.org/abs/2502.13592

188

DH Group at FBK @dh-fbk.bsky.social · 31/10/2025

@bsavoldi.bsky.social from @fbk-mt.bsky.social will present Translation in the Hands of Many: Centering Lay Users in Machine Translation Interactions at the poster session on Wed Nov. 5th, 11:00-12:30 in Hall C

072

DH Group at FBK @dh-fbk.bsky.social · 31/10/2025

@mrovera.bsky.social will present Multilingual vs Crosslingual Retrieval of Fact-Checked Claims: A Tale of Two Approaches at the poster session on Fri Nov. 7th, 10:30-12:00 in Hall C

120

DH Group at FBK @dh-fbk.bsky.social · 31/10/2025

@casulalm.bsky.social will present Job Unfair: An Investigation of Gender and Occupational Bias in Free-Form Text Completions by LLMs at the poster session on Wed Nov. 5th, 11:00-12:30 in Hall C

131

DH Group at FBK @dh-fbk.bsky.social · 31/10/2025

- Translation in the Hands of Many: Centering Lay Users in Machine Translation Interactions (preprint: arxiv.org/abs/2502.13780)

120

DH Group at FBK @dh-fbk.bsky.social · 31/10/2025

We'll be at EMNLP in Suzhou 🇨🇳 next week to present 3 main conf papers: - Job Unfair: An Investigation of Gender and Occupational Bias in Free-Form Text Completions by LLMs - Multilingual vs Crosslingual Retrieval of Fact-Checked Claims: A Tale of Two Approaches (preprint: arxiv.org/abs/2505.22118)

142

DH Group at FBK @dh-fbk.bsky.social · 24/10/2025

Next week we will be presenting our recent research at the LT highlights @ FBK day 2025! If you're interested, join us (and remember to register on Eventbrite!) lt-highlights.fbk.eu

lt-highlights.fbk.eu

LT Highlights @ FBK 2025

031

DH Group at FBK @dh-fbk.bsky.social · 02/10/2025

✨ Shared task alert! ✨ We are organizing FadeIT, the first shared task on fallacy detection accounting for genuine disagreement. FadeIT is part of EVALITA, whose workshop will be held in beautiful Bari, Italy in February 2026. Learn more from our website! 🌐 sites.google.com/fbk.eu/fadei...

Image showing the text "FadeIT: Fallacy Detection in Italian Social Media Texts, a Shared Task at EVALITA 2026".

054

Reposted by DH Group at FBK

AILC-NLP @ailc-nlp.bsky.social · 26/09/2025

Thanks to all the participants! #clicit2025

0115

Reposted by DH Group at FBK

AILC-NLP @ailc-nlp.bsky.social · 24/09/2025

First poster session at #clicit2025

022

Reposted by DH Group at FBK

Alan Ramponi @alanramponi.bsky.social · 25/02/2025

Happy to share that “Fine-grained Fallacy Detection with Human Label Variation” with @agnesedaff.bsky.social and @satonelli.bsky.social was accepted to #NAACL2025 main conference 🎉 📝 arxiv.org/abs/2502.13853 #NLProc #NLP #NAACL 1/🧵

Example showing multiple plausible span annotations provided by annotators A1 and A2 due to different interpretations for the text "American study: mutation spreads four times faster, but 💉 are needed" in Italian

1275

Reposted by DH Group at FBK

satonelli.bsky.social @satonelli.bsky.social · 06/02/2025

Last week we kick-started the TWIN4DEM project, where we plan to use #NLProc to study political discourse and executive aggrandisement. What a challenge! We definitely have three exciting years ahead of us 💪 #HorizonEurope @twin4dem.bsky.social

032

DH Group at FBK @dh-fbk.bsky.social · 17/12/2024

Wishing everybody happy holidays before a well-deserved break! Greetings from our traditional Christmas lunch 🎄🧑‍🎄 #NLP #Digitalhumanities

091

Reposted by DH Group at FBK

Marco Rovera @mrovera.bsky.social · 16/12/2024

Happy to kick off with good news: our paper "ModaFact: Multi-paradigm Evaluation for Joint Event Modality and Factuality Detection", co-authored with our intern Serena Cristoforetti and @satonelli.bsky.social , accepted at #COLING. Dataset+model soon available 💾💻, stay tuned! See you in Abu Dhabi!

052

Reposted by DH Group at FBK

land-fbk.bsky.social @land-fbk.bsky.social · 06/12/2024

Congrats to @daniel-russo.bsky.social for the Best Student Paper Award at #CLiCit2024! Congrats also to the co-authors Oscar Araque and @marcoguerini.bsky.social! 🥳

0195

Reposted by DH Group at FBK

Camilla Casula @casulalm.bsky.social · 07/12/2024

A bonus group pic 🌼

061

Reposted by DH Group at FBK

AILC-NLP @ailc-nlp.bsky.social · 06/12/2024

We start this new account with the group picture of #clicit2024! Follow us! #NLProc #NLP

0278

Reposted by DH Group at FBK

satonelli.bsky.social @satonelli.bsky.social · 05/12/2024

And here are today's oral presentations from @dh-fbk.bsky.social: NLP to extract gustatory language by Teresa Paccosi (now at @knaw-nl.bsky.social) and multiple models and datasets to classify documents with the EuroVoc thesaurus by Lorenzo Bocchi, @casulalm.bsky.social and @ziorufus.bsky.social

083

DH Group at FBK @dh-fbk.bsky.social · 03/12/2024

Hello Bluesky! We would like to kick off this account with the DH @ FBK starter pack, which you can use to follow the members of our research unit 🌟 go.bsky.app/5feLo44

095