A Development of Chatbot for Advising University Undergraduate Regulation

Abstract

This research aims to 1) develop a chatbot that provides advice on the regulations of Kasetsart University regarding undergraduate studies, 2) evaluate the effectiveness, and 3) assess satisfaction with the use of the chatbot. This research uses Microsoft Power Platform to develop a chatbot that answers questions about the regulations, such as how many credits can be registered in a regular semester or the qualifications for students to receive honors. The chatbot operates based on the following principles: 1) The chatbot receives search queries in sentences. 2) The search queries are tokenized using the PyThaiNLP API. 3) The tokenized queries are used to search in the FAQ database related to the regulations. 4) The relevant FAQs are displayed based on the search queries. 5) Users select the questions and answers they need. The FAQ database is created using FAQ generation technique by uploading the regulations into the FAQ generator system, which generates questions with answers and corrects and saves them into the database. After development, an effectiveness comparison is made between the answers generated by the proposed chatbot and those of the Copilot AI chatbot. The results showed that out of ten questions, the proposed chatbot was able to answer ten questions appropriately, while the Copilot AI chatbot was able to answer seven questions. Moreover, satisfaction assessment was conducted with a volunteer sampling of 61 first-year Computer Science students. The data were analyzed using descriptive statistics, which consist of mean and standard deviation. The assessment covered various aspects, including content, usability, design, and performance. The results showed that all aspects were rated as good, and the overall evaluation was also at a good level (Mean = 4.17, S.D. = 0.75)).