Developing AI chatbot framework in training regulation domain

Abstract

. In recent decades, AI chatbots (aka artificial intelligence virtual assistants) have been investigated and developed in numerous domains, such as e-commerce, health care, tourism, and education, because of their benefits in the interaction between humans and machines via natural language text/speech. Universities’ training regulation is necessary for students to plan their studies. However, students can only access these rules via the training handbook, their training counsellors, or the staff of the Office for undergraduate education. This process is time-consuming for students, lecturers, and office staff. To deal with this problem, we proposed an AI chatbot framework based on the Rasa platform and applied it to the university’s training regulation domain. The results show that the F-measure of the natural language understanding model is 84.6%, and the experts’ accuracy evaluation is 89.3%.