A personalized product recommendation method

Abstract

The invention relates to a personalized product recommendation method, comprising: receiving a user's product search instruction; if the user has been classified, calculating multi-dimensional average label information preferred by the user, and the user's recommendation model calculates the first multi-dimensional attribute based on the multi-dimensional average label information Information, calculate the distance between the first multi-dimensional attribute information and the second multi-dimensional attribute information of each product in the product category, and push the product with the smaller distance to the user; otherwise, based on the second multi-dimensional attribute information of multiple products, The first multi-dimensional average attribute information is calculated, and based on the first multi-dimensional average attribute information and recommendation models of various users, a recommendation model applicable to the user is determined. The recommendation method provided by the present invention firstly clusters users, and uses multi-dimensional product label information and product attribute information to make the recommendation accuracy high; in addition, when the user is a new user, through attention transfer learning, the existing recommended The model forms a recommendation model for new users to address the cold start problem.

Description

technical field [0001] The invention relates to the technical field of data mining, in particular to a method for recommending personalized commodities. Background technique [0002] With the continuous development of network technology, in today's era of information overload, both as information consumers and information producers have encountered great challenges: information consumers need to find the information they are interested in from the massive amount of information; Information producers need to make their information stand out and attract users' attention. The personalized recommendation method can solve the above problems very well. At present, the main personalized recommendation methods mainly include content-based recommendation methods and collaborative filtering recommendation methods. However, these methods generally have problems such as low accuracy and cold start. Contents of the invention [0003] The present invention provides a method for recomm...

AI Technical Summary

Problems solved by technology

[0003] The present invention provides a personalized product recommendation method to solve the problem of low recommendation accuracy existing in the prior art

Images