🚨 Happy to share some recently accepted papers from our group.
They cover a range of topics, from reasoning and hallucination mitigation to dialogue evaluation, bias, humour generation, and more 📚✨
If you’d like to take a look, the arXiv links are in the poster.
Funded PhD position available with me (and other supervisors) based in the School of Biosciences, looking at measuring impact of food systems chemicals on health.
@sheffieldnlp.bsky.social @shefcmi.bsky.social
findaphd.com/phds/project...
Does Multimodal Large Language Model Truly Unlearn? Stealthy MLLM Unlearning Attack
Xianren Zhang, Hui Liu, Delvin Ce Zhang, Xianfeng Tang, Qi He, Dongwon Lee, Suhang Wang
www.arxiv.org/abs/2506.17265
#EMNLP2025 Main
Beyond Hate Speech: NLP’s Challenges and Opportunities in Uncovering Dehumanizing Language
Hamidreza Saffari, Mohammadamin Shafiei, Hezhao Zhang, Lasana T. Harris, Nafise Sadat Moosavi
arxiv.org/abs/2402.13818
#EMNLP2025 Main
Comparing Apples to Oranges: A Dataset & Analysis of LLM Humour Understanding from Traditional Puns to Topical Jokes
Tyler Loakman, William Thorne, Chenghua Lin
arxiv.org/abs/2507.13335
#EMNLP2025 Findings
SheffieldNLP
PhD Project - ECOSOLUTIONS DFA - IMEMRE: Development of an impact measurement (Immune, Emergence, Resistance) for food systems chemicals on human immune disease, infection and AMR at University of She...
GreekBarBench: A Challenging Benchmark for Free-Text Legal Reasoning and Citations
Odysseas Chlapanis, Dimitrios Galanis, Nikos Aletras, Ion Androutsopoulos
arxiv.org/abs/2505.17267
#EMNLP2025 Findings
Multimodal Large Language Models (MLLMs) trained on massive data may memorize sensitive personal information and photos, posing serious privacy risks. To mitigate this, MLLM unlearning methods are pro...
How Private are Language Models in Abstractive Summarization?
Anthony Hughes, Ning Ma, Nikos Aletras
arxiv.org/abs/2412.12040
#EMNLP2025 Main
Enhancing Logical Reasoning in Language Models via Symbolically-Guided Monte Carlo Process Supervision
Xingwei Tan, Marco Valentino, Mahmud Elahi Akhter, Maria Liakata, Nikos Aletras
arxiv.org/abs/2505.20415
#EMNLP2025 Main
🎉 A warm welcome to Delvin Ce Zhang who recently joined us as a Lecturer!
His research applies Multi-Modal LLMs to real-world tasks like misinformation detection, recommender systems, and AI for science. Excited to have him on board and for many collaborations! 🌺🚀
sites.google.com/view/delvinc...
SheffieldNLP
Humour, as a complex language form, is derived from myriad aspects of life, whilst existing work on computational humour has focussed almost exclusively on short pun-based jokes. In this work, we inve...
arxiv.org
Dehumanization, i.e., denying human qualities to individuals or groups, is a particularly harmful form of hate speech that can normalize violence against marginalized communities. Despite advances in ...
We introduce GreekBarBench, a benchmark that evaluates LLMs on legal questions across five different legal areas from the Greek Bar exams, requiring citations to statutory articles and case facts. To ...
Great news!! Our scientific director Professor Kalina Bontcheva @sheffieldnlp.bsky.social at @sheffielduni.bsky.social has been appointed Chair of Working Group 1 for the new Code of Practice on the #Transparency of #AI -Generated Content.
www.veraai.eu/posts/kalina...
In sensitive domains such as medical and legal, protecting sensitive information is critical, with protective laws strictly prohibiting the disclosure of personal data. This poses challenges for shari...
arxiv.org
Large language models (LLMs) have shown promising performance in mathematical and logical reasoning benchmarks. However, recent studies have pointed to memorization, rather than generalization, as one...
