In order to fully tap and apply the information of textbook reviews on the e-commerce website, we use fine-grained emotional classification algorithm to analyze the user's online comments, based on the sentiment analysis results of product feature level, so as to assist customers and businesses to make reasonable and effective decision. In this article, we first use the crawler tool to collect online comment texts of teaching materials, and carry on some pretreatments such as denoising, segmentation and POS tagging, and then analyze the product features, based on the general emotional dictionary expands domain sentiment dictionary. Finally, based on the syntactic analysis results, combined with the language features of textbook comments, we design an affective tendency analysis algorithm which is suitable for the textbook reviews, and prove the validity of the algorithm through experiments.