Sentiment Classification on Twitter(X) Through Ensemble Deep Random Vector Functional Links

This study explores Twitter sentiment classification utilizing the Ensemble Deep Random Vectorial Functional Link (edRVFL), a non-iterative deep learning model. We approached sentiment analysis as a binary classification task, categorizing sentiments as either positive or negative within text data. Our datasets included 10,000 tweets from the 2017 Catalan refer-endum and 2,187 tweets from the 2010 Chilean earthquake. The edRVFL model integrates multiple layers of random vector enhancements to achieve deep representation learning, addressing the limitations of traditional single-layer networks. It also leverages ensemble learning to enhance generalization across different datasets, thereby improving both the robustness and accuracy of sentiment classification. Our findings suggest that the edRVFL model provides promising results in sentiment analysis, surpassing other models in metrics such as precision, recall, and accuracy. Future research will focus on validating the effectiveness of edRVFL across diverse datasets, enhancing its applicability in real-world sentiment analysis tasks.