معرفی
Sina Zarrieß is a Professor of Computational Linguistics at the University of Bielefeld, affiliated with the Faculty for Linguistics and Literature Studies. His research focuses on computational models of language use, with applications in natural language generation, dialogue systems, and language-vision integration. He leads the Computational Linguistics Bielefeld group and collaborates on interdisciplinary projects in ecology, hate speech analysis, and multimodal interaction.
Research interests include language model architecture optimization, pragmatic reasoning in generation, and ethical NLP. He actively contributes to workshops such as ACL, EMNLP, and NLP4Ecology, publishing on topics like character-based LMs, gaze-driven hate speech detection, and visualization-oriented dialog systems. His work bridges theoretical linguistics with applied machine learning, emphasizing interpretability and real-world impact.
Notable projects include developing BabyLM character-based models, studying gender-inclusive language benchmarks (SlayQA), and creating tools like VIST5 for adaptive visualization dialog. He explores how LMs encode linguistic principles like Maximize Presupposition! and investigates zero-shot learning for novel object categories.
Current research trends show a focus on model efficiency (e.g., sentence selection for classification), multimodal grounding (e.g., scene context in visual tasks), and human-AI interaction dynamics (e.g., explanation effects on user perception). Despite prolific publishing, no specific scientific awards are mentioned in the provided materials.
Advising and grants are not explicitly detailed, though his extensive co-authorship network indicates active mentoring. The Computational Linguistics Bielefeld group maintains open research directions in visual question answering, poetry generation diversity, and historical hate speech evolution studies.
