Mengxia Yu

Hi, I'm Mengxia Yu

I'm a PhD candidate in the Department of Computer Science and Engineering at the University of Notre Dame. I'm fortunate to be a member of Data Mining towards Decision Making (DM2 Lab), advised by Prof. Meng Jiang. My research interests are in NLP and LLM. Specifically, I focus on augmenting large language models with domain and task-specific knowledge for downstream scenarios, such as educational and scientific applications.

Prior to my PhD, I received my Bachelor's degree in Computational Linguistics from the Department of Chinese Language and Literature at Peking University.

News

  • October 2025: My work "Context Selection and Rewriting for Video-based Educational Question Generation" is accepted to EAAI 2026.
  • August 2025: My work "The Super Weight in LLMs"is reported by Apple Machine Learning Research.
  • June 2025: One co-authored paper QG-SMS on student modeling and simulation with LLM is accepted to ACL 2025 main. Congrats Bang!
  • May 2025: I am joining Amazon as an Applied Scientist Intern.
  • April 2025: Check out the preprint of my new work on Educational Question Generation.
  • Feb 2025: I passed belay test and now a certified top rope belayer.
  • Jan 2025: I passed my oral candidacy exam and now a PhD candidate!

Contact

Email: myu2 [at] nd.edu

Office: 355 Fitzpatrick Hall of Engineering

Location: University of Notre Dame, Notre Dame, IN 46556

Projects

Project One

COSER: Context Selection and Rewriting for Video-based Educational Question Generation

A new framework and dataset were created to improve the generation of educational questions from real-world lecture videos. The method combines context selection and rewriting strategies to generate more accurate and contextually relevant questions.

View Paper →
Project Two

The Super Weight in LLMs

We find that an extremely small subset of parameters in LLMs (in some cases, a single parameter) can exert a disproportionate influence on an LLM’s overall functionality. We try to uncover the mechanism behind this phenomenon.

View Paper →
Pre-training Language Models for Comparative Reasoning

Pre-training Language Models for Comparative Reasoning

This work introduces a new framework and data collection method for continual pre-training of LMs for comparative reasoning.

View Paper in EMNLP →
Project Four

Scientific Comparative Argument Generation

We created a new NLP task: generating comparative arguments that aim to present a scientific invention’s technical novelty by comparing it to one or multiple prior works.

View Paper in KDD DI →

In case you are curious...

Life Story of Me

October 1, 2025

My name is 虞梦夏, which means "dream of summer" in Chinese. I was born in the northwestern corner of Guangdong, China, as a proud member of the Zhuang people. I grew up in and consider myself a native of Foshan—a city you might know as the hometown of martial arts legends like Ip Man and Bruce Lee. While I'd love to say their epic skills rubbed off on me, my own kung fu abilities are, tragically, non-existent.

Where is University of Notre Dame?

September 15, 2025

University of Notre Dame is a private research university in South Bend, Indiana, United States. We are in the Midwest, yet we use Eastern Time.