Multi-relation collaborative filtering algorithm based on graph neural network

Abstract

The invention provides a multi-relation collaborative filtering algorithm based on a graph neural network, and the algorithm comprises the steps: 1, processing the historical interaction data of a user and an article, and extracting all user sequences UU and article sequences SI from the historical interaction data; and step 2, respectively constructing each user sequence SU and each article sequence SI into a user relationship graph GU = (VU, EU) and an article relationship graph GI = (VI, EI). The invention provides a multi-relation collaborative filtering algorithm based on a graph neural network. The article relationship and the user relationship are modeled at the same time by constructing the article relationship graph and the user relationship graph from the user and article interaction data, the multiple relationships are fused into the learning process of user and article interaction, and the influence of the multiple relationships on user and article interaction is learned, so that the model can better model user preferences.

Description

technical field [0001] The invention relates to the technical field of information processing, in particular to a multi-relational collaborative filtering algorithm based on a graph neural network. Background technique [0002] Collaborative filtering is one of the most widely used recommendation algorithms in the industry, which can effectively solve the problem of "information overload" caused by massive data. Collaborative filtering can learn user preferences based on historical interaction data (such as ratings, clicks) between users and items, so as to generate new recommended content for users. A good recommendation algorithm can not only help users find interesting content and improve user experience, but also increase traffic for merchants and create huge profits. [0003] However, traditional collaborative filtering models, such as matrix factorization and neural collaborative filtering, often only consider the relationship between users and items when modeling use...

AI Technical Summary

Problems solved by technology

[0003] However, traditional collaborative filtering models, such as matrix factorization and neural collaborative filtering, often only consider the relationship between users and items when modeling user preferences, while ignoring the relationship between items and users. , so that it is not enough to fully learn the user's preference from the historical interaction between the user and the item. The interaction between the user and the item is usually affected by two factors: 1. The user's historical preference, the interaction between the items that the user has interacted with in the past Relationships can directly affect a user's interest in a new item. Existing collaborative filtering algorithms that consider item relationships do not take into account the important role of user roles in the interaction between users and items; 2. User's social relationship, that is, user The relationship between items, a user's preference is often influenced by his social friends, which indirectly affects the user's decision-making behavior, and the existing recommendation models that consider user relationships also do not take into account the complex relationship between items, and need Clear social relationship data is often difficult to obtain in reality. In addition, traditional recommendation models do not consider that the relationship between different historical items (users) contributes differently to the model's prediction of user preferences.

Images