Literature recommendation method and system based on heterogeneous graph neural network

Abstract

The invention provides a literature recommendation method based on a heterogeneous graph neural network. The method comprises the following steps of obtaining user feature data and literature featuredata; extracting literature attribute data and literature citation data from the literature feature data, and obtaining literature citation feature data through a graph convolutional neural network according to the literature attribute data and the literature citation data; extracting structural data according to a preset meta-path, constructing a heterogeneous information network according to thestructural data, and obtaining structural feature data through a self-attention mechanism learning network; normalizing the source data to obtain a sample set, dividing the sample set into a trainingset and a test set, and inputting the training set and the test set into the recommendation model to train the recommendation model and obtain a recommendation result. Structural feature data of users and structural feature data of literatures are mined through a graph convolutional neural network and a graph attention network, diversified interest representation is carried out on the users, andthe purpose of carrying out personalized recommendation on the users is achieved.

Description

technical field [0001] The invention relates to the technical field of data recommendation, in particular to a document recommendation method and recommendation system based on a heterogeneous graph neural network. Background technique [0002] With the advent of the Internet era, information sharing presents a generalized and subtle development trend, and there are more and more scientific and technological documents. How to help users quickly find the content that users are interested in from a large number of scientific and technological documents has become a hot topic in the field of recommendation algorithms. question. Among the traditional recommendation methods, the most mainstream is collaborative filtering technology, which assumes that users are always interested in items that they were interested in in the past, and collaborative filtering technology can be classified into user-based collaborative filtering, item-based collaborative filtering, and user-based coll...

AI Technical Summary

Problems solved by technology

[0006] Based on this, it is necessary to address the technical problems that traditional heterogeneous neural networks and graph neural networks cannot be di

Images