Research on the Construction of User Portrait of Cross-border E-commerce Platform Based on Artificial Neural Network

With the vigorous development of global trade, the scale of cross-border e-commerce platform users is increasing, and the user needs and behavior characteristics are becoming more complex and diverse. Accurately constructing user portraits is of great significance for improving platform marketing effect, optimizing product recommendation and enhancing user experience. However, the traditional e-commerce user portrait construction method relies on statistical analysis, rule matching and other means, and there are many limitations in dealing with large-scale, high-dimensional and nonlinear user data. In this regard, this paper will explore the application feasibility of artificial neural network in this respect according to the actual application requirements, and propose a brand-new user portrait construction framework for cross-border e-commerce platforms, so as to achieve more accurate and efficient user portrait construction. Practice has proved that this framework is based on BP neural network and combined with gradient lifting decision tree (GBDT). Through comparative analysis, it is proved that this framework has practical advantages in user portrait prediction in the face of high-dimensional sparse features. At the same time, the horizontal comparative analysis experiment also shows the feasibility and practicability of BP neural network in constructing user portraits.