A Novel DTPB-EM Model for Evaluating GUI and Non-GUI Chatbots for Self-Directed Learning

Abstract

<p>Traditional learning models such as direct instruction learning, inquiry-based learning, problem-based learning, case-based learning, or self-directed learning are either teacher-centered or student-centered, and limit learners to learning from instruction materials, enquiry, group discussion, or internet searches for problem solving. Self-learning overcomes these limitations and uses a chatbot as a personal tutor for self-learning. The two major types of chatbot are GUI and non-GUI chatbots. The GUI chatbots require programming skills while others require no or minor skills for their creation that make learner difficult to select the appropriate chatbot to use. This research therefore aims to evaluate the strengths and limitations of existing GUI and non-GUI chatbots. We present a novel DTPB-EM model for evaluating the effectiveness of the chatbots for comparison, and conduct a face validity pilot test of the effectiveness of the use of GUI and non-GUI chatbots in education. The research methods applied here are review and expert evaluation. Roblox chatbot programmed with a GUI and a Non-GUI chatbots programmed with Python programming language were selected for comparison by five experts. It was found that the GUI chatbot gave positive results in regard to ease of use, compatibility, and self-efficacy, which implies that users will have a positive attitude towards its adoption. The non-GUI chatbot had slightly lower performance for these constructs, but gave good performance in resource and technology facilitating conditions in terms of response accuracy, database size, scalability and language support. This implies that a non-GUI chatbot offers good perceived behavior control. </p>