MUTutor: An LLM-based Support Tool for Teaching Assistants with Continuous Feedback in Asynchronous Q&A

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Final Year Thesis Oral Defense

Title: "MUTutor: An LLM-based Support Tool for Teaching Assistants with 
Continuous Feedback in Asynchronous Q&A"

by

GUO Bingcan

Abstract:

Asynchronous Questioning & Answering(Q&A) is a crucial resource for university 
students seeking assistance with lecture material, assignments, and course 
logistics from instructors and teaching assistants(TAs) beyond scheduled 
lecture hours. This interaction not only helps students but also allows TAs to 
develop skills in teaching and employing proper pedagogies, gain experience 
communicating with students, and deepen their grasp of the subject knowledge. 
Nonetheless, since TAs are often students who need to balance their research or 
coursework with TA responsibilities, many find it challenging to devote 
sufficient time to answer students' inquiries thoroughly and timely reflecting 
on their own TA skills.

To support TAs in efficiently responding to student questions and to facilitate 
an intuitive reflection on their TA skills, we develop MUTutor. This web-based 
system can provide specific insights on incoming questions and real-time 
feedback on TA's replies. The system harnesses the advanced capability of Large 
Language Models(LLMs), including text summarization, evaluation, and search 
functionalities, to optimize the asynchronous Q&A process for TAs in Computer 
Science 2(CS2) courses. Our usability testing with three novice teaching 
assistants for CS2 courses at HKUST indicates that MUTutors can effectively 
help them answer questions and receive constructive and personalized feedback 
to enhance their TA skills.


Date            : 29 April 2024 (Monday)

Time            : 14:00 - 14:40

Venue           : Room 4472 (near lifts 25/26), HKUST

Advisor         : Dr. MA Xiaojuan

2nd Reader      : Dr. TSOI Desmond Yau-Chat