Making Sales Strategies Based on the Existing Shopping Reviews
Jing-Cheng Shi, Yan Xiang Yang, Shi Qiu · Journal of Physics Conference Series · 2021
Abstract With the popularization of mobile Internet technology, people are increasingly inclined to choose to use the internet to purchase products. Consumers’ evaluation of products after online shopping will directly affect the future sales of the products, so many online shopping platforms are committed to the research of intelligent evaluation and analysis models, so as to improve their sales strategies in a timely manner. We take the evaluation of three kinds of goods on Amazon as the research object. The platform allows purchasers to give star-rating and text-based review of products, and also lets them vote for the helpfulness of the former reviews. Therefore, we need to figure out the relation between these factors. We construct an innovative Maximum-Relation-Minimum-Redundancy Feature Selecting Model to analyze how the text-based measures and ratings-based measures affect the reviews, especially on some occacsions at whichthe review has significant positive changes. We also discuss the connection between the specific quality descriptions of text-based reviews and the level of star-ratings. We use an innovative model, Bi-LSTM model to calculate the rate of different emotions contained in the review texts. Then we can find out a remarkably high correlation between certain descriptors and the star-ratings. This paper puts forward an intelligent and constructive scheme for e-commerce platform to adjust its own sales strategy based on customer’s reviews on products.