A Transfer-Based Deep Learning Model for Persian Emotion Classification

Sentiment analysis is the technical process of analyzing generated data to detect emotions. Emotion detection in text data involves identifying the sentiments expressed in the data. Most studies have focused on classifying opinions based on their positive or negative polarity in Persian text. In this study, a dataset of emotional sentences was prepared to label six basic emotional states, JAMFA. This dataset contains 2350 sentences and 31200 words. Several comparative models FastText-BiLSTM, CNN-BERT, RNN-BERT, and BERT-BiLSTM are performed to evaluate the results. Among all the proposed techniques, an optimized BERT-BiLSTM (OBB) strategy trains the model to focus on the most relevant class. The evaluation indicates that the accuracy of labeling is 92%, and the reliability of the dataset based on the type of emotion is 88%. The results show that the OBB model has achieved 86% accuracy in classifying basic emotions and reached 88% accuracy in binary classification.