Semantic Analysis on Product Review Headlines Based on Review Association Mechanism and Convolutional Neural Network

摘要

With the development of online shopping, product reviews and star ratings have an important impact on users' shopping behaviors, and the reviews can also urge merchants to improve their products. This article explores the internal connection between product reviews and star ratings, analyzes the key factors which deeply affect the star ratings, and constructs a review association mechanism based on high-frequency keywords to provide supplementary information for products with missing text reviews, which offers some valuable suggestions to merchants. At the same time, this article uses the headline which highly summarizes each product review to construct a convolutional neural network (CNN), and predicts the star ratings with high accuracy, which can also help users and merchants better understand those product reviews without star ratings.