Student satisfaction survey chatbot

Lee, Wei Jin (2024) Student satisfaction survey chatbot. Final Year Project, UTAR.

Preview

PDF Download (6Mb) | Preview

Abstract

Feedback is a piece of external information that is crucial for improvement. It can be found from different sources, whether it is a user of a product, a teacher guiding a student, or a customer in a restaurant. This project will be focusing on student’s satisfaction feedback about their university experience. Traditional web survey are widely used to collect feedback from students. University students tend to be more open to give feedback, and so, university management can take advantage of this to understand student’s problem. This project explores the implementation of AI chatbot in conversations with students and ask follow-up questions to gain insight on student’s university experience. Open-source libraries and models such as Natural Language ToolKit library, langchain and hugging face are used to integrate and modify the chatbot model and add functionality to it like predicting sentiment of a feedback. By leveraging on Natural Language Processing technology advancement, Large Language Models (LLMs) are used as the foundation for the text generation capabilities of the chatbot. To ensure high quality response, profanity filter and language detector models are also integrated using pre-existing python libraries like profanity-filter and langid. The result is a multifunctional chatbot system that can simultaneously predict sentiment of text, detect profanity, determine the language, and generate text.

Actions (login required)