Artificial intelligence and machine learning-powered personalized mathematics learning system for ordinary level students in Sri Lanka

Abstract

In today’s digital age, integrating technology into education has revolutionized how we learn, making complex subjects such as mathematics more accessible and engaging. Mathematics remains a significant challenge for Sri Lankan Ordinary Level (O/L) students, with high failure rates, limited personalized attention, traditional teaching methods, resource limitations, and language barriers. In response to these challenges, this study is focused on developing an artificial intelligence (AI) and machine learning (ML)-powered personalized mathematics learning system for grade 10 and grade 11 (O/L) students in Sri Lanka. The system is focused on addressing the individual learning needs and helping them improve math with diversified topics, incorporating a chatbot that offers immediate assistance to learners and a feedback mechanism embedded for checking the progress of the learners regularly. The study aims at developing ways of measuring the effectiveness of personalized learning paths in improving students’ mastery of specific mathematical concepts. Furthermore, it explores how AI and ML-based approaches correlate with pupils’ academic achievements by considering motivation levels in mathematics. Additionally, the study aims on developing tutoring functionalities within the system to foster improvements in student satisfaction and performance. Agile methodology was adopted for this study, enabling iterative development to improve the system throughout the project. The learning system underwent each development cycle, focusing on specific features and user feedback, which led to improvements in functionality and user experience. The system was mainly developed using the MERN stack, which comprises MongoDB for data storage, Express.js for server-side logic, React for the frontend, and Node.js for the backend. Furthermore, a linear regression machine learning model was trained for the real-time feedback system. Information on which topics students find challenging was gathered by surveying grade 10 and grade 11 students, both in person and online via Google Forms, with questions asking students to rate their difficulty level with various topics. In order to accommodate all the students, the surveys were conducted in both Sinhala and English languages, and the data gathered was used to train the chatbot using the RASA framework. The system evaluates students' basic math proficiency through a placement test and recommends a personalized learning package based on the test results. The system has shown significant improvements in math concept understanding and performance among 90% of ordinary-level students in Sri Lanka. The system's adaptability and real-time feedback, chatbot providing instant support, offer a viable solution to conventional teaching methods and improving education. Limitations include broader testing and scalability, with future work focusing on improving the AI model and feedback system. AI and MLpowered personalized systems are recommended to be used in schools to help each student learn better.