Donya Rooein
About Me
I am currently a postdoctoral fellow in the Department of Computer Science at Bocconi University. I am fortunate to work with Prof. Dirk Hovy at Milan Natural Language Processing Group (MilaNLP Lab). I joined MilaNLP LAB for INTERGRATOR project to study demographic factors to language processing systems and their role of them in the future of Conversational AI.
I'm on the academic and industry job market 2026. If you think I’d be a good fit to your team, I’d love to hear from you!
My research focuses on the intersection of Natural Language Processing (NLP) and Human-Computer Interaction (HCI), with a special emphasis on educational applications. I am particularly interested in evaluating Large Language Models (LLMs) in educational settings, with a focus on young users. My work includes developing metrics for text difficulty, creating benchmarks for complexity prediction, and examining cultural relevance and biases in AI-generated content for children. I am also interested in NLP for social good, and I work on studies related to biases and social norms in the Farsi language.
I received my Ph.D. in Information Technology Engineering at Politecnico di Milano, where I was fortunate to be advised by Prof. Barbara Pernici and Prof. Paolo Paolini. I was also a research visitor at ETH Zurich at my last year of PhD.
My Ph.D. thesis was “A Scalable, Reconfigurable, and Adaptive Framework for Chatbots in Education”. During My Ph.D., I focused on adaptive conversational agents; in particular, I designed and developed highly configurable chatbots in education to support various actors with different demographics.
Recent News
- January 2026: PATS: Personality-Aware Teaching Strategies with Large Language Model Tutors at EACL 2026
- January 2026: Can Reasoning Help Large Language Models Capture Human Annotator Disagreement? at EACL 2026
-
November 2025: Teacher Demonstrations in a BabyLM’s Zone of Proximal Development for Contingent Multi-Turn Interaction at 🏆 BabyLM EMNLP 2025
-
November 2025: Biased tales: Cultural and topic bias in generating children’s stories at EMNLP 2025
-
November 2025: Co-detect: Collaborative discovery of edge cases in text classification at EMNLP 2025
- September 2026: Exploring Subjective Tasks in Farsi: A Survey Analysis and Evaluation of Language Models) , Arxiv
-
July 2025: [Measuring Gender Bias in Language Models in Farsi](https://aclanthology.org/2025.bea-1.44.pdf](https://aclanthology.org/2025.gebnlp-1.21.pdf) at GeBNLP, ACL 2025
-
July 2025: [Large Language Models for Education: Understanding the Needs of Stakeholders, Current Capabilities and the Path Forward]([https://aclanthology.org/2025.bea-1.44.pdf at BEA, ACL 2025
-
July 2025: Educators’ Perceptions of Large Language Models as Tutors: Comparing Human and AI Tutors in a Blind Text-only Setting at BEA, ACL 2025
-
July 2025: Are Large Language Models for Education Reliable for All Languages? at BEA, ACL 2025
-
April 2025: Can I Introduce My Boyfriend to My Grandmother? Evaluating Large Language Models’ Capabilities on Iranian Social Norm Classification at NAACL 2025.
- April 2025: Invited talk at Università della Svizzera italiana (USI) on Child-Centred AI in Education.
- November 2024 – February 2025: Visiting researcher at ETH Zurich, Language, Reasoning and Education (LRE) Lab.
Selected Projects
📚 PATS: Personality-Aware Teaching Strategies with Large Language Model Tutors
A taxonomy of teaching strategies was developed based on learning science literature and LLM-based evaluations of different teaching strategies.
🔗 View Demo
📚 Biased Tales
A dataset designed to analyze how biases influence protagonists’ attributes and story elements in LLM-generated stories.
🔗 View Demo
⚖️ LegalBot
A collaborative project between Politecnico di Milano and Tribunale di Milano aimed at improving access to legal information. LegalBot uses a chatbot interface to assist users in understanding common legal concepts.
🔗 View Demo
🧑🏫 TalkyTutor
A chatbot designed to support educators and students through adaptive, chatbot-mediated learning. Built using the aCHAT framework, TalkyTutor empowers non-technical users (e.g., teachers) to customize both content and conversation flow.
Publications
Please find all publications on my Google Scholar.