Research on the Algorithm of Big Data Technology in E-Commerce User Portrait Construction

This paper proposes an e-commerce user portrait construction algorithm based on big data technology, combining deep learning and cluster analysis, and realizing accurate portrait construction through multi-dimensional user behavior data and social data. The algorithm first uses a deep learning model (LSTM) to extract the long-term behavior characteristics of users, and then combines the K-means clustering algorithm to group users, and finally generates personalized user portraits. Experimental results show that the proposed algorithm shows significant advantages in the construction of user portraits on e-commerce platforms. Compared with traditional rule-based algorithms and other common clustering methods, this algorithm improves the accuracy of user portraits by 18% and the computational efficiency by 30 %. In addition, the effectiveness of the algorithm in personalized recommendation and precision marketing is verified through practical application cases, providing e-commerce platforms with more accurate user analysis tools. This study provides an innovative user portrait construction method for e-commerce platforms, and improves the application effect of e-commerce platforms in user behavior prediction and personalized recommendation.