Sentiment Prediction Based on Deep Learning for Intelligent E-Learning Systems

Sentiment analysis on feedback information plays a crucial role in intelligent e-learning systems (IES). To enhance customer service quality through reviews, predicting sentiment polarity is crucial. This paper focuses on integrating sentiment analysis into an online programming course in intelligent e-learning systems to better understand the challenges faced by students. E-learning platforms, typically web-based, are vital for reflecting users’ experiences, emotions, and purchase intentions. Recently, sentiment analysis systems have frequently been used on popular platforms like Facebook and Twitter, helping companies identify potential customers across different segments. Utilizing sentiment analysis assists in making informed product adjustments to meet user satisfaction. Our approach employs modern frameworks to build deep learning models, specifically leveraging the pre-trained RoBERTa model. We integrate this into a website application with a Django back-end, facilitating an easy implementation of RoBERTa into an API that communicates with Next.js, a cutting-edge framework for full-stack developers. This combination ensures optimal performance for the user experience and scalability of the project. Experimental results demonstrate that incorporating sentiment analysis into websites significantly enhances trust among users in e-learning systems.