Welcome to my personal website!

I am a postgraduate student at the University of Sydney, majoring in Data Science with a specialisation in Machine Learning.

I have recently embarked on my research journey in the world of Natural Language Processing (NLP). Here are my research interests:

• Trustworthy LLMs: Using my interdisciplinary grounding in social sciences and NLP, I hope to explore ways to improve the reasoning, efficiency, and robustness of LLMs in the same way that other insights borrowed from fields such as neuroscience and physics have inspired the development of AI technologies.
• Dialogue Detection: Conversations occur in various contexts. LLMs serve as invaluable assistants in identifying negative factors and offering effective strategies during discussions, thereby highlighting the significant value of LLMs.
• Theory Simulation: Just like Stanford’s Smallville, using LLMs as agents for social simulation and theory discovery can not only validate and discover existing sociological theories, but also reflect the capabilities and limitations of LLMs themselves.

Currently, I am delving deeper into research and actively seeking opportunities for research internships (One of the reasons is that I have no supervisor or mentor so far) and pursuing a PhD or MS/PhD.

Finally, I would like to thank who have accompanied and helped me in my life and work, especially my senior, Hongbin Na and Zimu Wang, for their patient and professional advice and pleasant collaboration.

Degrees

• M.S. in Data Science, University of Sydney, 2025.07 (expected)
  • GPA: 3.22/4.00 (current)
• B.S. in Economics, University of Sydney, 2023.12
  • GPA: 3.85/4.00

Papers

(The asterisk * appearing after authors means equal contribution.)
(OSS means One-Sentence-Summary about the paper.)

• [COLING2025] [2024] Detecting Conversational Mental Manipulation with Intent-Aware Prompting
  • Author: Jiayuan Ma*, Hongbin Na*, Zimu Wang, Yining Hua, Yue Liu, Wei Wang, Ling Chen
  • Award: Best Short Paper, Oral
  • OSS: Based on Theory of Mind as motivation, a new approach, Intent-Aware Prompting (IAP), was introduced to detect mental manipulation in dialogues, which notably reduces false negatives and improving overall performance on detection.
• [CLPsych2025] [2025] From Posts to Timelines: Modeling Mental Health Dynamics from Social Media Timelines with Hybrid LLMs
  • Author: Zimu Wang*, Hongbin Na*, Rena Gao, Jiayuan Ma, Yining Hua, Ling Chen, Wei Wang
  • Award: Ranked 3rd in the Shared Task
  • OSS: This work proposes a hybrid large language model framework to model individual mental health dynamics by first identifying adaptive and maladaptive evidence in social media posts, and then synthetically generating user well-being scores as well as post-level and timeline-level summaries.

Internships

• 2023.12 - 2024.02, ESG Research Intern
  • PwC China (Shanghai, China)
• 2023.06 - 2023.08, VC Research Intern
  • Shenzhen Capital Group Co. Ltd (Shanghai, China)
• 2022.09 - 2022.12, IBD Research Intern
  • China Securities Co. Ltd (Beijing, China)
• 2021.06 - 2021.07, Consulting Assistant Intern
  • Capgemini Consulting (China) Co. Ltd (Hangzhou, China)

Skills

• Computing
  • Python, R, STATA, MS Office
• Languages
  • Chinese, English, Spanish (B1), Japanese (N5)
• Certificates
  • ACCA (passed 5/13), CFA Level 1 (in preparation)