UTARGpt

Loh, Jun Xiang (2024) UTARGpt. Final Year Project, UTAR.

Preview

PDF Download (4Mb) | Preview

Abstract

The UTARGPT project introduces a highly innovative solution in the form of a chatbot designed specifically to serve the students and staff of UTAR by centralizing scattered university-related information into a single, accessible point. This significantly simplifies the process of accessing vital university details, which are currently distributed across numerous online platforms, making it difficult for users to find the information they need quickly and efficiently. To streamline this access, we employ a sophisticated web crawler to gather the latest updates and information from the UTAR websites. This web crawler is an essential component of the system as it automates the collection and assembly of data into PDF documents. These documents are subsequently uploaded into a GPT-based knowledge base, which the chatbot utilizes to provide responses to user inquiries. Ensuring the chatbot remains accurate and up-to-date with the latest information is critical. To manage this, a cron job scheduler has been integrated into the system. This scheduler regularly triggers the web crawler to perform data updates, thereby maintaining the chatbot's relevancy and effectiveness in delivering the most current information available to its users. Furthermore, the UTARGPT chatbot is designed to offer a personalized experience for each user. This personalized approach enhances the user experience by making the information not only more accessible but also more applicable to each student's unique situation. In addition to UTARGPT, I will develop OpenAI Assistant chat integrated with RAG, . This parallel development allows the team to compare the performance and user experiences between these two chatbot. Through this comparative analysis, insights gained will be used to continually refine and enhance both platforms. The development of these two different systems ensures a broad evaluation of potential solutions, aiming to optimize the way students and staff interact with university information, ultimately improving the overall efficiency and user satisfaction.

Actions (login required)