Cross-domain recommendation method fusing generative adversarial network with double generators and double discriminators and auto-encoder

Abstract

The invention provides a cross-domain recommendation method of a generative adversarial network and an auto-encoder fusing double generators and double discriminators. The method comprises the following steps: acquiring two same-type data domains with the same project or the same user, and taking the same user information or project information in the two data domains as auxiliary information; cascading the score data in the single data domain with user information or project information serving as auxiliary information; carrying out feature extraction on the data obtained after cascading; aligning and fusing the features between different data domains through the confrontation process of the generative adversarial network to obtain fused data features; and finally, decoding the fused data features to obtain score prediction matrixes of the two data domains, and recommending items with relatively high prediction scores to the user through the score prediction matrixes.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence recommendation, in particular to a cross-domain recommendation method that combines a generative confrontation network with dual generators and dual discriminators and an autoencoder. Background technique [0002] With the continuous development of network technology, people can obtain more and more data. However, massive amounts of data can make it difficult for users to find the information they need. Therefore, in order to solve this problem, the recommendation system came into being. However, recommender systems usually face the problems of data sparsity and cold start, and cross-domain recommender systems provide a new method to solve the problems of data sparsity and cold start. [0003] In reality, there are usually the same items or users between shopping websites or video websites. Usually, due to the fact that the sparsity of the target domain is higher than th...

AI Technical Summary

Problems solved by technology

Usually, since the sparsity of the target domain is higher than that of the source domain, resulting in high data heterogeneity between the source domain and the target domain, performing bidirectional migration will lead to negative transfer.

Therefore, the single-objective cross-domain recommendation system is difficult to improve the recommendation effect of the target domain and the source domain at the same time, and does not make full use of the data of the target domain and the source domain.

Images