Shanshan Xu

📄 Check out the preprint: arxiv.org/pdf/2510.08524

We propose a prompt optimization framework that combines Monte Carlo Tree Search with a lightweight proxy prompt evaluator to efficiently discover better prompts for unfair ToS clause classification.
🧵2/2

🎉 Proud moment: my students’ paper get accepted at @nllpworkshop.bsky.social @EMNLP 2025
💪 so proud of the students' dedication and growth throughout the project
🚩 Also a milestone for me - my first last-author paper
🌱 It’s been a new and interesting experience: part author, part reviewer 🧵1/2

📣New adventure! 🇩🇰Just joined Uni Copenhagen as a postdoc with @danielhers.bsky.social. I'll be working on LegalNLP & human-centered AI🤗 Excited for the research, collaborations, and much hygge & lykke✨fun fact: I stumbled on the book lying on the street in Munich just before applying the position 🔮⭐️

I'll present our work w/ @santosh-tyss.bsky.social
@yanai.bsky.social @barbaraplank.bsky.social on LLMs memorization of distributions of political leanings in their pretraining data! Catch us at L2M2 workshop @l2m2workshop.bsky.social #ACL2025 tmrw
📆 Aug 1, 14:00–15:30 📑 arxiv.org/pdf/2502.18282

This list is definitely incomplete, so feel free to share more relevant work!
While it’s great to see increasing engagement , I notice the terminology varies widely, 💬 human label variation, 💡perspectivism in NLP, and 📊 pluralistic human values 🤔 which is sometimes a bit confusing to me.

👏 Upcoming workshop: Perspectivist Approaches to NLP @EMNLP 2025
Focusing on non-aggregated datasets and multi-perspective modeling, with sessions on labeling, modeling, evaluation, and applications.
nlperspectives.di.unito.it

NLP for Positive Impact Workshop Overview

🔊 @vinodkpg.bsky.social – Invited Talk #NLP4PosImpact #ACL2025: “AI for the World of Many”
- Also a shout-out to the three talks above that inspired this thread!
sites.google.com/view/nlp4pos...

🗣️ Lora Aroyo – Keynote #NeurIPS2023: “The Many Faces of Responsible AI”

- callling for recognizing the inherent ambiguity of content and the natural diversity of human perspectives, urging to "treat label disagreement as signal, not noise".
neurips.cc/virtual/2023...

🔈 @barbaraplank.bsky.social – Keynote #ACL2024: Are LLMs Narrowing Our Horizon? Let’s Embrace Variation in NLP!
- Introducing human label variation across the ML pipeline, and arguing that "embracing human label variation" is crucial for building trustworthy NLP
2024.aclweb.org/program/keyn...

📢 @verenarieser.bsky.social – Keynote #ACL2025: “Whose Gold? Re-imagining Alignment for Truly Beneficial AI”
- Discusseing the ethical and technical challenges of aligning AI when human values conflict, and proposes a three-level alignment pyramid as a path forward.
2025.aclweb.org/program/keyn...

Super excited to see the growing interest and awareness in diversity, variation, and pluralistic alignment in the NLP community! ✨ 🪩

Below 🧵 are some keynote talks and workshops that highlight this important direction 🤸

I'll present our work w/ @santosh-tyss.bsky.social
@yanai.bsky.social @barbaraplank.bsky.social on LLMs memorization of distributions of political leanings in their pretraining data! Catch us at L2M2 workshop @l2m2workshop.bsky.social #ACL2025 tmrw
📆 Aug 1, 14:00–15:30 📑 arxiv.org/pdf/2502.18282

Congrats! 🎊🎊

Dei Boarisch heard ned bei "Servus" und "Pfiade" auf? Dann suach ma genau Di!
Wir suachan Bairischsprecher:innen, de a kurze Umfrage über KI-generierds Boarisch für a Masterarbeit beantwortn mechadn.
Mid jeder Teilnahme bring ma den boarischn Dialekt a Stickal weida in de digitale Weyd!

Are you attending NAACL 2025 and are you interested in low-resource languages and dialects?

Then don't miss our very own @verenablaschke.bsky.social's keynote talk at the WNUT 2025 workshop on May 3rd:

Beyond “noisy” text: How (and why) to process dialect data

🌐 ☀️
noisy-text.github.io/2025/

Congratulations Dr. @LijunLyu for a fantastic PhD defense 👏🎓🎉

Catch me in oral talk on split voted case and human-model alignment in 🪷loutus suit 5-7🪷16:00⌚Session: Interpretability and model analysis (15 minutes to go) #ACL2024

Poster session now!📜How aligned is the model 🤖with human 👀on the split voted cases, which are often social - politically controversial? Come to visit us at poster 1⃣2⃣4⃣ in CCA1 #ACL2024

📜Poster: 14:00, Mon. 12.Aug
📢Oral: 16:00, Tue 13. Aug

Please drop by if you are interested in human label variation, pluralistic human values and NLP for LegalTech. Looking forward to insightful discussion. (2/2)

✈️ On my way to #ACL2024 in Bangkok🇹🇭 I'll present our work 🔍Through the Lens of Split Vote: Exploring Disagreement, Difficulty and Calibration in Legal Case Outcome Classification 🔍with @TYSSSantosh2 Oana @barbara_plank @matgrabmair (1/2🧵)

📰 https://dir.lat/EzcxUr

What can AI/ML researchers learn from 🙋survey methodology to make data collection 🎯 less biased and more 😀human centric? @stephnie @barbara_plank and @fraukolos are presenting their position paper in hall C #2007! Go and see it! #ICML2024

Almost there ...... today (In 90 minutes) ⏲️5:40 pm, catch me in talk on 📢human disagreement and 📏model calibration at 🧑‍⚖️ Legal Tech Social 🪩in 🎵Lehar1-4🎶 #icml2024 #legaltech #NLProc (Joint work with @TYSSSantosh2 , Oana, @matgrabmair, @barbara_plank )

🚀 Heading to 🇦🇹 #ICML2024! I’ll give talk about pluralistic human values and alignment at the Machine Learning Scientists in Legal Tech session tomorrow (Wed.) afternoon. If you're interested in human label variation and NLP for LegalTech, please reach out! 👋

I really enjoyed the poster session today at #emnlp23. Thank you all for stopping by 😀 (Interestingly, we all seem to agree on the human disagreement) @mumblamb @barbara_plank @TYSSSantosh2 @LeonStaufer

Happening now! Come to the Law Law Land in Aquarius 2 for 🐤BoF session of 🪶 NLP on legal text ⚖️ #EMNLP2023

⏲️ 14:00 - 15:30 join us at the🦜BoF 🪶session at Aquarius 2 📢No matter you are an NLP researcher on #LLM or a lawyer with an #LLM feel free to join everything under Law, Language and LLM 3/3

Food Republic @ VivoCity · 1 HarbourFront Walk, #03 - 01,... ★★★★☆ · Food court

⏲️12:30 Meet us at the poster if you want some street food and Singapore cuisine 😋🇸🇬🍲. We will go for lunch at 大食代/Food Republic in VivoCity 2/3 https://maps.app.goo.gl/5ZX7zsCStLRxfs6B7

TUM LegalTech Rundown today (8. Dec)@EMNLP23: 11- 12:30 We’ll present our poster“From Dissonance to Insights: Dissecting Disagreements ...............” at East Foyer. Come and stop by to chat about Human Label Variation, Disagreement and LegalNLP 1/3🧵

📢Check out the preprint🔗https://arxiv.org/abs/2310.11878
👍Many thanks to the amazing interdisciplinary team (NLP, Legal, Linguistics) and multi-institution collaboration (TUM, LMU, IHEID and RUB) 🎉! @TYSSSantosh2 @matgrabmair @barbara_plank Oana Ichim @IsabellaRisini 6/6

Our results highlight the intricate nature of case outcome classification and emphasize the necessity for interdisciplinary collaboration between legal experts and ML researchers to integrate expertise into trustworthy and explainable COC system aligned with legal principles. 5/6