The Workshop on Natural User-generated Text (WNUT) will be at EMNLP this year (noisy-text.github.io/2026/), with a shared task on lexical normalization with a focus on Asian languages (noisy-text.github.io/2026/mu…)
I am hiring: candidate.hr-manager.net/App… So if you are interested in training/evaluating LMs and language/linguistics, consider applying!
Wondering how "distant" or "diverse" your languages are?, need different perspectives on distance? DistaLs v0.1.1 is now out!
Rob van der Goot retweeted
📢 Looking for a PhD student to start in Sep 2024! Applications due April 15
This project is about LLM competence in interpreting prompts and responding to them:
candidate.hr-manager.net/App…
If you're at @eaclmeeting, you can catch @robvanderg who will co-supervise this project!
Workshop on Noisy and User-Generated Text (WNUT) is collocated with EACL (@eacmeeting) this year, Looking forward to your submissions!,
submission deadline: 18th of December
Rob van der Goot retweeted
Introducing... the Subspace Chronicles (ft. @robvanderg, @barbara_plank, @iatitov)—a riveting tale of linguistic information emerging, shifting and disentangling across LM training! Coming to an #EMNLP2023 near you! 📜⏳
📑personads.me/x/emnlp-2023-pa…
📽personads.me/x/emnlp-2023-vi…
ALT Subspace Chronicles via probes θ across LM training time, as measured by F1, codelength L, and subspace angles across tasks and time.
Updated my website, and now finally you can reply to my "blogs" (robvanderg.github.io/blog/)! Looking forward to see if anyone cares enough to start a discussion there!
📜 Paper alert: arxiv.org/abs/2304.13989 📜: we (Kia Kirstein Hansen, I'm just the supervisor), investigate transfer from WSJ to Elder Scrolls data with fine-grained POS tagging. A domain adaptation setup where the writing style is similar, but the topics are wildly different 1/3
Interestingly, the most difficult cases are proper nouns and adjectives, where even humans are struggling. Altough the data was too small to get it through peer review, the analysis section is an excellent example on how to do qualitative analysis. 2/3
Rob van der Goot retweeted
More info on RAD-day (Feb 13, 11am-3pm CET) talks is now available at itu-dasyalab.github.io/RAD/r….
Thanks Rob van der Goot @robvanderg, Zoi Kaoudi, Matthias Boehm @matthiasboehm7, Florina Ciorba @cflorina, Luca Maria Aiello @lajello, Oana Balmau @oanabalmau for being part of this.
Rob van der Goot retweeted
On-going now: VarDial@EACL 2023 shared tasks, incl. slot & intent detection for low-res lang varieties (SID4LR)
@noeminaepli Yves Scherrer @robvanderg & myself
VarDial Tommi Jauhiainen @nljubesic @TiedemannJoerg @preslav_nakov @marcos_zampieri
🥳📢 sites.google.com/view/vardia…
Rob van der Goot retweeted
Full room for @robvanderg's talk in our research seminar today! It's been great to have him back with us for a week, and to hear what he's been up to since he left Groningen: a lot of cool research! Thanks Rob, and come back soon. #NLProc @FacultyofArtsUG
Rob van der Goot retweeted
I’m happy to share that our paper “Tafsir Dataset: A Novel Multi-Task Benchmark for Named Entity Recognition and Topic Modeling in Classical Arabic Literature” has been accepted to appear at #COLING2022 as an oral presentation!
#MachineLearning #NLProc #LowResourceLanguages
For any master students expecting to graduate this year, please consider applying for this PhD position!, if you have any questions about the position or scope do not hesitate to contact me. Deadline 14-03, link: candidate.hr-manager.net/App…
Also bored by shared tasks where the largest finetuned transformer model wins on a classification task?, consider joining this one!
To nrmlze or n't to nrmlze? 🤔
If you're interested in lexical normalization: at this year's WNUT 2021 there is a shared task on lexical normalization, including data from 12 languages (+code-switched data): noisy-text.github.io/2021/mu… - organization led by @robvanderg
Ok, fair enough, a large transformer model still ranked highest. But they had to do some interesting things to get there! Congrats to David Samuel and @straka_milan (aclanthology.org/2021.wnut-1…) as well as all other participants!, many interesting approaches were used.
The overview paper is here: robvanderg.github.io/doc/Mul… , and will be in the anthology soon