Research on recommendation strategy of e-commerce user portrait based on user dynamic interest factor hybrid recommendation algorithm

Abstract

In modern society with the good and fast development of mobile e-commerce, the commodity information and the user behavior data accompanying it show an explosive and sudden growth trend, which also leads to the emergence of information overload on the e-commerce platform, and the proposed personalized recommendation system for e-commerce users largely alleviates this problem mentioned above. The personalized recommendation system for e-commerce users aims to solve the information overload of e-commerce platform by analyzing the user behavior data of e-commerce platform, so as to explore the interest preference of e-commerce platform users and make active recommendation of advertising content related to e-commerce platform. Although the research on recommendation algorithms for e-commerce platforms has made great progress, there are still challenges in terms of sparse data, static user features and interpretability of e-commerce platform recommendation results in terms of big data feature recognition. Therefore, in this paper, a hybrid recommendation algorithm based on the forgetting curve of e-commerce platform and the automatic feature construction of e-commerce platform is studied in the e-commerce scenario of e-commerce platform, combined with the e-commerce data collected in real field, for the sparsity of e-commerce platform data, the interpretability of recommendation results and the static nature of e-commerce platform user features.