Enhancing Sentiment Analysis Accuracy on IMDB Reviews Through Ensemble Machine Learning Techniques

In the rapidly evolving field of sentiment analysis, the IMDB movie review dataset has become one of the key benchmarks for evaluating the performance of machine learning models. This paper presents a comprehensive study of various machine learning models applied to the dataset, focusing on positive and negative reviews. We delve into the intricacies of advanced models such as BERT, LSTM, and GRU. The novelty of this work lies in the application of ensemble methods, specifically voting functions and stacking, to improve the accuracy of sentiment classification. We propose original techniques that leverage the strengths of individual models, mitigating their weaknesses through a collaborative approach. The ensemble methods used outperform single-model approaches, demonstrating improved accuracies in the classification of both positive and negative reviews. This research contributes to the ongoing discourse in sentiment analysis, offering fresh perspectives and techniques that enhance sentiment classification accuracy. The findings underscore the potential of ensemble methods in machine learning.