Abstract:
"In response to the evolving landscape of education within the digital era, this research explores the critical challenge of understanding sentiments expressed by students in online learning environments. The transformation towards e-learning, notably accelerated during the COVID-19 pandemic, necessitates a deeper exploration of student feedback as it is an essential source for improving modern teaching techniques. Even though there is a profound shift towards online learning in the Sri Lankan education system, the substantial hindrance of a specialised system in analysing the student's sentiments poses a significant barrier. It recognised the importance of sentiment analysis and building a system that could fit the unique linguistic nuances of the Sri Lankan education domain, enhancing the overall e-learning experience.The provided research methodology unfolds as a systematic, largely data-driven approach that can address the identified problem comprehensively. From a thorough literature review to the reutilization of existing techniques, a dual-language sentiment analysis system that integrates a lexicon-based approach and machine learning models was created. Also, the implemented system emphasises using a rule-based approach to enhance domain relevancy. It compares its efficiency with the labelled data while providing novel insights into student sentiments within the Sri Lankan education context. Preliminary results highlight the innovative nature of the developed sentiment
analysis system, showcasing its effectiveness in navigating the complexities of multilingual
education feedback. Quantitative evaluation metrics, including accuracy, affirm the system's robust performance. This research provides a foundation for a more refined learning experience and opens avenues for future work in sentiment analysis within the diverse linguistic landscape of online education. "
