Aspect Based Sentiment Analysis on Amazon Book Review Using DistilBERT

Product reviews are valuable sources of information in e-commerce, yet existing sentiment analysis models struggle to capture fine-grained opinions on specific product aspects, especially for books. This study introduces an improved Aspect-Based Sentiment Analysis (ABSA) model tailored for Amazon book reviews, leveraging the DistilBERT model for natural language processing. The model is designed to classify sentiments related to specific book aspects such as plot, character development, and writing style, achieving high accuracy, precision, recall, and F1 scores. While the model excels in distinguishing positive and negative sentiments, challenges remain in accurately classifying neutral sentiments. To further enhance the model, future research will explore advanced fine-tuning techniques, incorporate diverse feedback sources, and integrate multimodal data. Addressing class imbalance and ethical considerations such as bias in sentiment classification will also be critical for improving the model’s robustness and application in real-world scenarios.