Advancing Human-Computer Interaction: A Stacking Classifier Approach to Textual Sentiment Analysis using Ensemble Machine Learning

The integration of sentiment analysis in textual content analysis improves human-computer interactions in various dimensions. This research work aims to advance text-based emotion detection systems, with a particular emphasis on hidden emotional identification thorough textual analysis. This research study proposed a Stacking Classifier by integrating machine learning algorithms: logistic regression, linear support vector classification (linear SVC), multi-layer perceptron (MLP) classifier, random forest, and decision tree. This ensemble generates probabilities for individual emotions, which are then fed into a meta-classifier (Logistic Regression) for accurate emotion identification. Empirical findings from the ISEAR dataset, which includes four separate emotion categories, demonstrates the efficiency of proposed Stacking Classifier model in textual content analysis. Significantly, the classifier attains a higher testing accuracy of 89% and a testing accuracy of 90%. These results indicate the potential of proposed ensemble methodology in advancing the field of text-based emotion detection.