Ontology and large language model based intelligent tutoring system

Abstract

The evolution of online education systems traces back to distance education methodologies of the 18th and 19th centuries, which laid the foundation for independent learning. The late 20th century saw the convergence of technological breakthroughs, including personal computers and the internet, leading to the development of computer-assisted instruction (CAI) and later, learning management systems (LMS). Online education systems evolved to embrace asynchronous learning, multimedia capabilities, and social learning principles, experiencing further growth during the COVID-19 pandemic. Prominent online learning platforms like Coursera, Udemy, and Skillshare have revolutionized access to education. these platforms have limitations regarding course quality consistency, instructor engagement, and accessibility. Artificial intelligence (AI) is transforming online education through Intelligent Tutoring Systems (ITS) and Expert Systems (ES). ITSs provide personalized and adaptive learning experiences, while ES assists with decision-making and problem-solving tasks. However, existing ITSs face limitations in scalability, automation, and feedback provision. In this research, the hypothesis posits that an ITS can be developed effectively using Ontology and LLMs. This concept is inspired by the unique capabilities of Ontology in structuring human knowledge and LLM in possessing strong Natural Language Processing (NLP) capability and general knowledge. The proposed system generates teaching plans, resources, assessments, quizzes, grades, progress tracking, and personalized feedback. Teaching plans outline schedules and topics, while resources enhance learning. Assessments and quizzes evaluate comprehension, with grades providing insights. Progress tracking identifies strengths and weaknesses, facilitating targeted interventions. Personalized feedback guides improvement strategies. The system is divided into four main components which are the domain knowledge model, tutoring model, student model, and user interface. These components are designed to be segregated, allowing modifications to one component without affecting the others. the evaluation of the system is done using the focused approach and comparative analysis. Using the focused approach find out if the combination of ontology and LLM is solving the issues that can’t be solved by ontology and LLM working individually in the system. Using comparative analysis find out the system is better than the selected similar systems for the experiment.