A brand recommendation method, electronic device, storage medium and system

Abstract

The present invention provides a brand recommendation method, comprising: obtaining several order data, combining the order data of the same user name to obtain user order data, using the user order data and user name as training data; inputting the training data into a preset recommendation model In the method, the training data in the preset recommendation model is trained by using the logistic regression algorithm and the random negative sampling algorithm and the trained recommendation model is obtained; the list of active users on the online shopping platform is obtained, and the data of the brands sold on the online shopping platform are obtained; Input the list of active users and the data of brands on sale into the trained recommendation model for matching and get a list of recommended brands. A brand recommendation method of the present invention solves the problem that there was no way to completely customize the list of brands they are interested in independently for each user in the past, and at the same time uses the training model for recommendation matching throughout the whole process, which increases the accuracy and efficiency of recommendation and improves the user experience. sense of experience.

Description

Technical field [0001] The present invention relates to the field of data processing, and more particularly to a brand recommendation method, an electronic device, a storage medium, and a system. Background technique [0002] Since the current online shopping platform has a large number of brand goods every day, users have only seen a limited number of brands every time, how to make users interested in the limited number of brands have become the focus of research. [0003] It is currently using two ways: 1. Based on business personnel to sensitive to goods and users, manually determine brand sorting; To predict the click rate of different brands to determine sorting. However, when using two ways, as the population is classified, the workload of manual sorting has increased, and the characteristics of each user group are determined, and it is also necessary to consume a lot of energy, and all users in another user group see the brand or The same, there is no way to customize the ...

AI Technical Summary

Problems solved by technology

However, when the above two methods are used, with the increase of crowd classification, the workload of manual sorting will increase sharply, and it will take a lot of energy to determine the characteristics of each user group. The brand seen by all users in the other user group is still Similarly, there is no way to completely customize the list of brands they are interested in independently for each user

Images