With the vigorous development of e-commerce, user-generated content (UGC) has become a core resource for insight into consumer needs in the e-commerce ecosystem. Sentiment analysis technology, through natural language processing and machine learning methods, identifies of emotional tendencies in text data such as reviews and bullet screens, providing support for consumer decision-making, merchant operations, and platform optimization. This paper adopts the methods of literature review and case study to systematically sort out the technological evolution of sentiment analysis in the e-commerce field, including typical models in the stages of traditional methods, machine learning, and deep learning. Combined with segmented scenarios such as fresh produce, home appliances, and pre-prepared dishes, it analyzes its application practices in topic mining and sentiment classification. Finally, it discusses challenges such as data noise and multimodal fusion in technical applications. It looks forward to future trends such as large language model fine-tuning and cross-domain adaptation, providing a reference for sentiment analysis research and applications in the e-commerce field.
Y Shen (Tue,) studied this question.