Explainable Sentiment Analysis Dataset

We utilize two widely used sentiment analysis datasets: Amazon Reviews, a multi-class dataset with 1–5 star ratings categorized into five sentiment levels, and IMDB Movie Reviews, a binary classification dataset distinguishing between Positive and Negative sentiments. While most sentiment analysis datasets focus on broad sentiment categories, fine-grained classification is crucial for advancing research in this field. To address this, we developed a high-quality, fine-grained sentiment dataset with the help of human volunteers. This dataset includes 1,838 Amazon reviews and 1,447 IMDB movie reviews, split into 70% training and 30% test sets, facilitating more nuanced sentiment analysis.