Advancing Emotion Detection in a Text of Transformer-Based Models and Traditional Classifiers
S. Jayanthi, S. S. Arumugam · 2024
Detection of emotions from text poses a formidable and intricate challenge within the domain of text analysis. Practitioners in the realm of sentiment analysis are directing their efforts towards developing applications for emotion recognition, prompted by the escalating interest among virtual communities, individual users, and commercial entities in comprehending and aggregating public opinions. Nevertheless, a multitude of prior research endeavors focusing on emotion detection have leaned on less efficient machine learning classifiers and limited datasets, resulting in a decrease in overall performance. In order to address this concern, the current study evaluates the efficacy of sophisticated transformer-based models, specifically RoBERTa and EmoBERTa, utilizing a standardized emotion corpus. The outcomes of the conducted experiments demonstrate the potential of these models in relation to various assessment metrics like precision, recall, and F1-score. Ultimately, the investigation identifies and recommends the most appropriate model for emotion classification based on its performance.