Bingsheng Arthur Yao
I am an associate research scientist in the Khoury College of Computer Sciences at Northeastern University (PI: Prof. Dakuo Wang).
My research at the intersection of NLP and HCI focuses on human-centered design of NLP applications to support collaborative work in domain-specific scenarios, including medical, healthcare, and education. Recently, I am working on LLM role-playing agents for human behavior simulation and human-agent collaboration.
Our lab is looking for self-motivated research assistants, Ph.D. students, and postdocs. Please refer to our lab website for more information.
Research
Patient Care
A critical sector with pressing needs that involves care provider teams, caregivers, and patients. A considerable amount of my recent work focuses on the human-centered design of AI/LLM systems to support stakeholder collaboration for remote patient monitoring (RPM) and clinical decision-makings, include identifying needs and challenges encountered by diverse stakeholders, designing and evaluating AI/LLM systems with these stakeholders.
Sepsis [CHI ‘24]; older adults [IMWUT ‘24a], post-operative patients [1] [2] [3] [CHI ‘25], concussion [4]
Children Education
Another unique collaborative scenario of my HCI & NLP research. I worked with stakeholders (i.e., education experts and parents) to understand their painpoints in storytelling activities, curate high-quality Question-Answering (QA) datasets, develop domain-specific NLP models, and design AI-supported interactive storytelling systems.
FairytaleQA [ACL ‘22a] & StorySparkQA [EMNLP ‘24] datasets, QA Generation Models [ACL ‘22b], StoryBuddy system [CHI ‘22]
Enhance AI With Human Cognitive Behaviors
This summarizes my pursuit of the long-term research vision. AI could outperform humans in task-solving performance, however, humans’ rational cognitive behaviors form real intelligence. My research on low-resource learning techniques pave the path for personalized AI when we provide individualized rationales as additional signals. Lately, I am devoted to advance LLM Role-Playing Agents to realistically mimic human behaviors, simulate human social dynamics, and establish human-agent collaborations.
Active Learning [EMNLP ‘23], In-Context Sampling [NAACL ‘24], Mental-LLM [IMWUT ‘24b], LLM Role-Playing Agents [ACL ‘25]
Appointment
- Associate Research Scientist
2025 - Current, Northeastern University - Postdoc Research Associate
2024 - 2025, Northeastern University
Education
- Ph.D. in Computer Science
2019 - 2024, Rensselaer Polytechnic Institute (Advisor. Prof. Jim Hendler) - M.S. in Information Technology
2018 - 2019, Rensselaer Polytechnic Institute - B.S. in Computer Engineering
2014 - 2018, Rensselaer Polytechnic Institute
Note
Please refer to my Google Scholar page for the most up-to-date publication record.
The best way to reach out is through emails: b [dot] yao [at] northeastern [dot] edu.
News
| 2025.02 | Co-first authored paper Survey of LLM Role-Playing Agent Evaluation is now publically available on arXiv |
|---|---|
| 2025.01 | Five papers were accepted to CHI 2025. Thanks for the hard working by collaborators and mentees! |
| 2024.10 | Our paper StorySparkQA Dataset with Real-World Knowledge for Children Education was accepted to EMNLP 2024 |
| 2024.09 | Our paper Secret Use of Large Language Models was accepted to CSCW 2025 |
| 2024.07 | Our paper Early Sepsis Prediction with Uncertainty Quantification and Active Sensing was accepted to KDD 2024 |
| 2024.04 | Our paper LLM-based Voice Assistant for Asynchronous Older Adults-Care Provider Communication and Mental-LLM were accepted to IMWUT 2024 |
| 2024.03 | First-authored paper In-Context Sampling Strategy for Reliable LLM Prompting was accepted to NAACL 2024 Findings |
| 2024.03 | Guest talk at USC titled “Bridging AI Research and Real-world Scenarios”. Thanks Prof. Yao Du for the invitation! |
| 2024.02 | I am joining Prof. Dakuo Wang’s Human-Centered AI Lab at Northeastern University as a postdoc associate! |
| 2024.01 | Two of our papers, Human-AI Collaboration in Sepsis Diagnosis and User’s Sensitive Disclosure with LLM were accepted to CHI 2024 |
| 2024.01 | I passed the Ph.D. dissertation defense. My deepest gratitude to all those who supported and helped me, especially Prof. Jim Hendler and Prof. Dakuo Wang |
| 2023.10 | Our paper, Discourse Framework for Science Journalism, was accepted to EMNLP 2023, and another first-authored paper, Active Learning Empowered by Natural Language Explanations, was accepted to EMNLP 2023 Findings |
| 2023.07 | First-authored paper Objective Evaluation of Human Explanations was accepted to ACL 2023 for Oral Presentation |
| 2022.05 | Two first-authored papers, QA-Pair Generation for Story Books and FairytaleQA Dataset were accepted to ACL 2022 |
| 2022.04 | Our paper StoryBuddy was accepted to CHI 2022 |
| 2021.09 | Our paper Narrative Open-Domain QA Techniques was accepted to TACL (2021) 9 |
| 2020.03 | Our paper Trust in AutoML was accepted to IUI 2020 |