A Detailed Approach to Persian Sentiment Analysis Using Fine-Tuning Strategies

In recent years, sentiment analysis has gained substantial attention due to its application in various fields such asmarketing, social media monitoring, and customer feedback analysis. This paper presents a comprehensive approach for fine-tuning the ParsBERT model, a transformerbased architecture specifically designed for Persian text, to perform sentiment analysis. We used a constructed dataset from user comments, implement a systematic methodology for model training, and evaluate the model's performance. Our experiments demonstrate the effectiveness of fine-tuning ParsBERT in understanding nuanced sentiment expressions in Persian, yielding a competitive performance compared to existing models.