Recommendation method for interested commodities of user

Abstract

The invention discloses a recommendation method for interested commodities of a user. According to the method, a user-commodity historical consumption behavior is learnt from a user and a commodity historical consumption behavior by using a multi-pair sorting algorithm; commodity preference matrix and user-to-user model using matrix factorization model The commodity preference matrix is modeled, the hidden vector representation information of the users and the commodities is fully considered, and the commodities potentially interested by each user can be adaptively mined through the adaptive sampling method, so that the recommendation effect of the multi-pair ranking recommendation method is more accurate.

Description

technical field [0001] The invention relates to the fields of machine learning and recommendation systems, in particular to a method for recommending products of interest to users. Background technique [0002] Collaborative filtering algorithm is one of the most popular algorithms in recommender systems. Previous research has paid more attention to the collaborative filtering algorithm based on user rating data, but in more application scenarios in life, it is difficult for us to obtain user ratings on products. The historical records of a large number of users do not contain explicit rating information (such as the products purchased by the user, the URLs followed), and we call this type of data the user's implicit feedback. In recent years, scholars from various countries have gradually shifted their research attention from explicit feedback to implicit feedback recommendation. Different from the explicit feedback data, for each user, the implicit feedback data only con...

AI Technical Summary

Problems solved by technology

Such a sampling method has two disadvantages: first, it is a static sampling method that lacks adaptability; second, it is a global sampling method that lacks individuality

The above two shortcomings affect the recommendation ability of the multi-pair sorting algorithm and limit the learning and understanding of the multi-pair sorting algorithm for data.

Images