Comment text-oriented graph neural network recommendation method

Abstract

The invention discloses a comment text-oriented graph neural network recommendation method, which comprises the following steps of: 1, constructing a bipartite graph of a user and a commodity by using comments and scores of the user on the product, including a user node set, a product node set, a score matrix and a comment feature tensor; 2, constructing a comment text-oriented graph convolution method, taking comments and scores as edge features to participate in graph convolution, and coding user and product characterization; and 3, enhancing the characterization of the user and the product by utilizing graph contrast learning. 4, constructing an interaction layer, and coding an interaction vector from the representation of the user and the product; and 5, drawing close the distribution of the interactive representation and the comment vector through comparative learning. And 6, predicting a score according to the interactive characterization so as to realize product recommendation. According to the method, comments and scores are used as interaction features of the user and the product, the mutual influence between the comments and the scores is adaptively learned, more accurate node characterization and finer user preference are learned, and therefore the recommendation performance can be improved.

Description

technical field [0001] The present invention is designed in the field of personalized recommendation, specifically, a graph neural network recommendation method oriented to comment text. Background technique [0002] With the rapid development of Internet technology, a large amount of information has emerged in the network service platform for users to choose, and the problem of information overload has become increasingly prominent. Personalized recommendation system recommends products that users are interested in by analyzing user's historical behavior and mining user's interest preferences. A large amount of user comment information in e-commerce websites can help to further explore user preferences and improve the recommendation effect. Current mainstream methods using review text treat historical reviews as descriptive texts of users and products, and extract semantic features from them to enhance the representation of users and products. [0003] Although existing r...

AI Technical Summary

Problems solved by technology

[0003] Although existing review-based recommendation methods achieve more accurate recommendation results than traditional recommendation models, these methods ignore that user-product interaction can be modeled naturally with a bipartite graph: users and products are nodes, ratings and review text Reflects the characteristics of the edges between nodes

From the perspective of bipartite graphs, the existing comment-based recommendation methods directly use the comment text on the side as the node attribute, which inevitably introduces noise into the node representation learning, making it difficult for the node representation to reflect the real user preferences and product features, thus limiting recommended effect

Images