Cross-Border E-commerce Sales Prediction Model Based on Deep Learning

Abstract

Cross-border e-commerce can be considered an efficient strategy for businesses seeking speedy growth, especially in the wake of the information technology revolution and further growth of the Internet economy. In this study, a model is proposed that combines the use of Convolutional Neural Networks (CNN) with an attention mechanism in encoding and selecting product image features. At first, a five-layer CNN is built without the fully connected layer to extract information from an image effectively. The attention mechanism is applied to identify the elements of an image which most significantly influence the generation of relevant textual descriptions over time. Furthermore, it transforms the customer evaluation process to price perception through the analysis based on quantitative indicators of the pricing model providing a mathematical correspondence to enhance understanding and analysis.