Emotion Analysis of English-Translated Tamil Literature

Focus of this study is to investigate the nuanced world of emotions in English-translated Tamil literature through the lens of sentiment analysis. The dataset consists of poems expressing a range of emotions including joy, sadness, love, anger, pride, courage, and fear. To effectively analyze and classify these emotions, we utilized cutting-edge transformer-based models, like BERT, DistilBERT, and RoBERTa, to conduct our analysis. The outcomes of our study showcase the exceptional capabilities of these models in effectively analyzing emotions within poetic expressions. RoBERTa, in particular, emerged as a standout choice, consistently achieving high accuracy, precision, recall, and $F 1$ scores across all emotion categories. Its extensive pre-training and dynamic masking techniques allowed itto excel in capturing the intricate emotional nuancesembedded in the translated poems. DistilBERT demonstrated commendable efficiency without compromising accuracy, making it a practical option for real-time sentiment analysis. BERT, although a strong performer, showcased marginally reduced efficiency compared to RoBERTa and DistilBERT.