Educational Platform in Higher Education With LLM-Driven Chatbot for Computer Networks and Telecommunications Course

Abstract

Large language models (LLMs) brought the possibility of enhancing chatbot interactions in educational platforms by providing students adapted support throughout the learning process. This study presents the design and development process of an LLM-powered educational platform prototype and its evaluation in a single learning session on the Computer Networks and Telecommunications course at a technical faculty specialized in information systems and technologies. The proposed platform combines a microlearning paradigm and the availability of LLM chatbot support after each learning unit for students to clarify misconceptions and deepen knowledge and comprehension. The prototype was built using Angular for the client-side application and Express.js for the backend with Firebase Firestore as the database solution. It leveraged the OPENAI GPT-4o model through the available Application Programming Interface (API) for providing the chatbot, which was instructed to offer support for students in the context of the mentioned course. As a result, students generally found the chatbot to be a useful support tool in the learning process, with higher effectiveness reported for clarifying theoretical concepts compared to assisting with practical tasks. This perception aligns with expert evaluations of the chatbot's responses, which revealed statistically significant differences between theoretical and practical answers in terms of factual accuracy, relevance, completeness, and coherence, but not in fluency. Despite these observed differences, both student and expert assessments were overall positive across both types of questions. Even though single-session evaluation limits generalizability of results, these findings suggest feasibility and promising results within this educational context since the chatbot is perceived as a valuable component of the learning experience.