Applying Machine Learning to Enhance the Accuracy of Text-Based Sentiment Analysis

Sentiment analysis serves as a powerful tool for tracking public opinion across sectors like marketing, politics, and social media. Traditional classification techniques often struggle with contextual ambiguities, sarcasm, and linguistic variations. To address these challenges, this research proposes an ensemble sentiment analysis model combining Support Vector Classifier (SVC) and Random Forest Classifier (RFC) with sentiment lexicon-based features. Through TF-IDF-based feature extraction and a voting ensemble strategy, the proposed system improves classification performance. Experimental evaluation using the IMDB movie review dataset demonstrates that the ensemble model achieves an accuracy of 90.2%, precision of 88.5%, recall of 87.9%, and F1-score of 88.2%, outperforming standalone classifiers. Lexicon integration enhances semantic interpretation, particularly for neutral and context-dependent statements. The model’s robustness and adaptability make it suitable for real-world applications like social media monitoring and customer feedback analysis.