Research

My research interests are at the intersection of learning sciences, educational data science, and algorithmic fairness. I use data and modeling to explore how we learn and collaborate, while identifying and addressing biases that may arise in these processes.

🤖 Learning Analytics for Human-AI Interaction

The rise of generative AI, particularly Large Language Models (LLMs), holds great potential to make education more accessible and personalized. Data collected from these generative AI powered platforms can reveal insights into how we interact with AI, and under what conditions successful learning can occur. My goal is to analyze data from these platforms using machine learning and natural language processing to understand more about how students learn and interact.

As part of the UC Irvine Language and Learning Analytics lab, I work with Dr. Nia Nixon to develop a platform for Human-AI Teaming that simulates various interpersonal dynamics within team settings. By customizing AI personas, we can examine how team dynamics influence group productivity and individual learning outcomes, including situations that would be challenging or even harmful to study in real-world experiments. Specifically, I am interested in how students from historically marginalized groups experience team dynamics differently.

Also as part of the UC Irvine Digital Learning lab guided by Dr. Mark Warschauer, I work on developing PapyrusAI, an LLM-powered writing platform designed to scaffold the writing process. Using data collected from this platform, I am building a learning analytics dashboard that effectively analyzes and visualizes student-LLM interactions, furthering our understanding of how AI can support learning and engagement in writing tasks.

🔍 Algorithmic Bias and Fairness in Education

As algorithms become more central to educational tool development and practices, it is essential to recognize that they are not inherently value-neutral. Every algorithm reflects the choices of its designers and developers – applying them without considering bias risks reinforcing the very inequities that students from historically marginalized communities already face.

My work aims to identify and reduce bias in educational algorithms. Under the guidance of Dr. Shamya Karumbaiah, I have been evaluating commonly used algorithmic bias mitigation strategies in the context of education, examining their limitations and exploring alternatives. I am particularly interested in how societal biases intertwine with statistical biases, potentially amplifying inequities.

  • Choi, J., Karumbaiah, S., Matayoshi, J. (2025) Bias or Insufficient Sample Size? Improving Reliable Estimation of Algorithmic Bias for Minority Groups. Proceedings of the 15th International Learning Analytics and Knowledge Conference (ACM LAK). [Accepted]

🛠️ Tools for Learning Analytics Research

One of my research areas focuses on developing computational tools for learning analytics research. While many tools exist, they are often highly data-driven and emphasize quantitative data exclusively, leaving rich qualitative data collected from interviews or field studies overlooked.

At UW-Madison, under the guidance of Dr. David Williamson Shaffer, I designed and developed tools that bridge qualitative and quantitative methods – a framework known as Quantitative Ethnography, such as Epistemic Network Analysis (ENA) and nCoder. Specifically, I led the design and development of an automated qualitative coding tool, nCoder+ (now called Codey). This tool aims to increase the transparency of automated qualitative coding processes while minimizing human coding effort for validation.

  • Choi, J., Ruis, A. R., Cai, Z., Eagan, B., & Shaffer, D. W. (2023). Does Active Learning Reduce Human Coding?: A Systematic Comparison of Neural Network with nCoder. Fourth International Conference on Quantitative Ethnography 2022. [Nominated for the Best Paper Award]

  • Choi, J., Tan, Y., Eagan, B., & Shaffer, D.W. (May, 2023). Visualizing a Qualitative Data with the Epistemic Network Analysis (ENA). Research Analytics Series Workshop, Prince of Sonkla University, Thailand (Virtual).

  • Choi, J. & Cai, Z. (November, 2022). Introduction to Codey: Better, Faster, Fairer Qualitative Coding. Workshop hosted by the Penn Center for Learning Analytics (Virtual).

  • Choi, J., Cai, Z., & Lee, S.B. (October, 2022). Introduction to Automated Coding: Codey. Workshop hosted by the Fourth International Conference on Quantitative Ethnography (ICQE22).

  • Choi, J., Puetz, B., & Siebert-Evenstone, A. (October, 2021). Introduction to Automated Coding: nCoder. Workshop hosted by the Third International Conference on Quantitative Ethnography (ICQE21).