Research On Product Recommendation Based On User Portraits

With the rapid development of the e-commerce industry,smart recommendation technology has gradually become a research hotspot in the field of e-commerce.Due to the accelerating pace of life,people have higher and higher requirements for shopping experience.Therefore,it is eagerly hoped that the system can recommend suitable products for them according to their shopping habits in order to save shopping time and cost.However,it is challenging to realize personalized recommendation of products by mining users' interest characteristics.Personalized Product Recommendation provides great convenience for the majority of shoppers.In this paper,based on the user log data of the mall as the analysis data set,combined with user portrait,deep learning,natural language processing,factorization machine model,recommendation algorithm and other technologies,the research of mall recommendation system based on user profile is carried out.This paper first combines the user profile technology with the characteristics of the e-commerce industry,designs the user profile label and data collection index by determining the user profile dimension,and then uses the triple tuple to model the user profile,and realizes the generation and management of the user profile based on the tag visualization technology.Secondly,a recommendation algorithm based on user profile and factor decomposition machine is proposed.The multi-dimensional features extracted from user profile information are introduced into the factor decomposition machine,which makes the feature extraction of discrete information better,and effectively solves the problem of data sparsity and improves the recommendation effect.Through the verification and analysis on the real data set,the validity and rationality of the UP-FM recommendation algorithm proposed in this paper is proved,and the recommendation quality can be significantly improved.