publication 1
Towards a Comprehensive AI Teaching Assistant Based on Course Forum
Authors: Bowen Gu, Hao Wang, and Kaizhuo Chen
Published on: Carolina Digital Repository
In this paper, we present the kernels of a system based on natural language processing. Together with future implementation of the database, receiving and sending module, and user interface, the system can be served as an intelligent teaching assistant for educational forum posts processing for small courses, which aims at alleviating the workload of instructors. Currently, this system is capable of conducting the following tasks: (a) Automatically label new posts into pre-designed categories. (b) Identify posts whose content is similar or identical to another existing post. (c) Identify posts that do not contain enough information for the instructors to understand or answer the question. (d) Automatically answer students’ questions about the implementation of coding assignments using English translation of the solution code. We applied six language models, four prediction models, with two API packages to build the system.
The full paper is available at the Carolina Digital Repository.
A detailed illustration video (47 mins) is available here.
A concise illustration video (4 mins) is available here.