Sentiment Analysis of Movie Reviews Using Deep Learning

The rapid proliferation of mobile devices has driven the integration of artificial intelligence (AI) technologies, making the efficient operation of complex deep learning models within resource-limited environments a critical challenge in AI. This study addresses this challenge by focusing on mobile applications and proposing a streamlined sentiment analysis model specifically tailored for English movie reviews. We begin by exploring the diverse applications of sentiment analysis in user feedback and public opinion monitoring, underscoring its growing significance. Subsequently, we provide an overview of the evolving landscape of related technologies. In our experiment, we utilize the bag-of-words model for data preprocessing, which we combine with a convolutional neural network (CNN) and dropout regularization to construct a lightweight model architecture. Through iterative parameter optimization, we successfully achieve a delicate balance between model performance and complexity. We trained the model on a dataset consisting of 20,000 samples and achieved an accuracy of 81.76% on a test set of 5,000 samples. Compared to traditional, more complex sentiment analysis models, our proposed approach significantly reduces computational demands, demonstrating its potential for practical deployment on mobile devices.