YA-TA: Towards Personalized Question-Answering Teaching Assistants using Instructor-Student Dual Retrieval-augmented Knowledge Fusion

TL;DR

The paper tackles scalable, personalized tutoring in large classes by introducing YA-TA, a virtual teaching assistant grounded in instructor lectures and tailored to individual students. It presents the DRaKe framework (Dual Retrieval-augmented Knowledge Fusion) that concurrently retrieves instructor-side knowledge $K_I$ and student-side knowledge $K_S$, then fuses them in a reasoning process to produce responses $r_t = f(D_t, K_I, K_S)$. Evaluations via G-Eval and qualitative case studies on CS50 and Yonsei courses show that dual retrieval enables better alignment with both instructor philosophy and student understanding, with extensions like a Q&A Board and Self-Practice enriching learning. The work highlights practical implications for reducing TA workload while maintaining pedagogical fidelity and personalized support in real classroom settings.

Abstract

Engagement between instructors and students plays a crucial role in enhancing students'academic performance. However, instructors often struggle to provide timely and personalized support in large classes. To address this challenge, we propose a novel Virtual Teaching Assistant (VTA) named YA-TA, designed to offer responses to students that are grounded in lectures and are easy to understand. To facilitate YA-TA, we introduce the Dual Retrieval-augmented Knowledge Fusion (DRAKE) framework, which incorporates dual retrieval of instructor and student knowledge and knowledge fusion for tailored response generation. Experiments conducted in real-world classroom settings demonstrate that the DRAKE framework excels in aligning responses with knowledge retrieved from both instructor and student sides. Furthermore, we offer additional extensions of YA-TA, such as a Q&A board and self-practice tools to enhance the overall learning experience. Our video is publicly available.

Figures (5)

• Figure 1: The motivating example of YA-TA. A typical LLM faces challenges in providing responses that consider both instructor and student sides. YA-TA addresses these issues by employing a DRaKe framework.
• Figure 2: The overview architecture of YA-TA. The image of the response is a screenshot of the YA-TA user interface. In YA-TA's final response, the part highlighted in blue indicates where instructor-side personalization is evident, while the part highlighted in orange indicates areas where student-side personalization is evident.
• Figure 3: Extension of YA-TA
• Figure 4: Virtual student profiles that we set
• Figure 5: G-Eval prompt used to assess precision