A Survey and Analysis of Sentiment Classification Using Machine Learning Algorithms

Sentiment analysis has become a crucial task in natural language processing for extracting meaningful insights from textual data. This research proposes a hybrid model using optimization techniques to enhance the accuracy and efficiency of sentiment classification. The proposed framework begins with a comprehensive preprocessing phase, including data cleaning, tokenization, stop-word removal, and normalization to improve data quality. Feature selection is performed using a combination of Genetic Algorithm optimization and TF-IDF representation to identify the most relevant features and reduce dimensionality. The refined feature set is then fed into multiple base classifiers, namely Naïve Bayes, Decision Tree, and Support Vector Machine (SVM), to capture diverse learning patterns. To further improve predictive performance, an ensemble model employing voting and bagging techniques is utilized to combine the outputs of individual classifiers. This integrated approach ensures robustness and reduces model variance. The system classifies sentiments into positive, negative, and neutral categories, followed by a detailed result analysis using evaluation metrics such as accuracy, precision, recall, and ROC curve. Experimental results demonstrate that the proposed hybrid model significantly outperforms traditional standalone classifiers in terms of classification accuracy and reliability. This study highlights the effectiveness of combining optimization techniques with ensemble learning for improved sentiment analysis performance.