Commodity personalized recommendation method and device, equipment, medium and product

Abstract

The invention relates to a commodity personalized recommendation method and device, equipment, a medium and a product in the technical field of e-commerce information, and the method comprises the steps: obtaining personalized information of a user, which comprises personal feature information, commodity preference feature information and time-space feature information; using different neural network models to respectively determine a first popularity feature corresponding to the personalized feature information, a second popularity feature corresponding to commodity feature information of candidate commodities, and a third popularity feature corresponding to comprehensive feature information formed by personalized information and the commodity feature information; intervening third popularity feature calculation according to the first and second popularity features to obtain effective popular popularity; and determining whether to push the candidate commodities according to the effective popularity. According to the method, the popularity of the commodities in the commodity database can be better restored, the interested commodities better conforming to personal preferences of the users can be matched for the users, more potential long-tail commodities have the opportunity to be exposed, and the Mattai effect in the commodity recommendation process is reduced.

Description

technical field [0001] The present application relates to the technical field of e-commerce information, and in particular, to a method for personalized recommendation of commodities and a corresponding device, computer equipment, computer-readable storage medium, and computer program products. Background technique [0002] Commodity sorting algorithms have proven their worth in various fields, and users are increasingly relying on the system for commodity recommendation. At the same time, the system itself will have a series of biases, such as exposure bias, selection bias and popularity bias. [0003] For the popularity bias, it is defined as popular items becoming more and more popular, which is caused by the current training paradigm, and for users, it is also the source of the user's "information cocoon room". The existence of popularity bias will reduce the degree of personalization, reduce the fairness of recommendation and exacerbate the Matthew effect. There are t...

AI Technical Summary

Problems solved by technology

[0007] However, the above methods only focus on how to adjust the weight of long-tail products, lack of consideration of how popularity affects each specific interaction, and lack of systematic understanding of the popularity deviation mechanism, so their actual performance in practice is limited , so how to eliminate the information cocoon room of product recommendation, so that more long-tail products have the opportunity to be discovered, there is still room for exploration and improvement of related technologies

Images