This study aims to develop an emotional state classification system for text, focusing on poetry and formal text, areas which have received limited attention in computational intelligence research. Leveraging Deep Learning techniques, specifically attention-based models, including Random Forest, Decision Tree, and Support Vector Classification (SVC), the objective is to classify text into distinct sentiments, such as positive or negative. By addressing this gap in emotional state classification across diverse textual genres, the research contributes to advancing the application of advanced techniques in sentiment analysis.