Opinion Mining and Sentiment Analysis of User Comments on E-Commerce Platforms—Taking Toner Products as an Example

在线评论作为消费者对产品和服务的主观表达，具有重要的研究价值。本文以电商平台“天猫”的爽肤水商品的初评与追评为研究对象，基于UIE模型对评论进行观点提取，计算语义相似度并构建商品特征体系，最后根据数据探究商品评论的关注度与满意度的分布规律。结果表明：用户在初评时对使用体验最为关注，在追评中用户对适用肤质和服务与物流的关注度提升较大；追评的满意度普遍低于初评，其中包装与设计特征的满意度降幅最大，达到12.3%。本文为商家和平台提供了产品的优化建议，为电商行业的健康发展提供一定参考。Online comments, as subjective expressions of consumers towards products and services, have important research value. This article takes the initial and follow-up reviews of toner products on the e-commerce platform “Tmall” as the research object. Based on the UIE model, opinions are extracted from the reviews, semantic similarity is calculated, and a product feature system is constructed. Finally, the distribution rules of attention and satisfaction of product reviews are explored based on the data. The results showed that users were most concerned about the user experience during the initial review, and in the follow-up review, users showed a significant increase in their attention to suitable skin types, services, and logistics; The satisfaction with the follow-up evaluation is generally lower than that of the initial evaluation, with the packaging and design features showing the largest decrease in satisfaction, reaching 12.3%. This article provides product optimization suggestions for merchants and platforms, providing a certain reference for the healthy development of the e-commerce industry.