Unlock instant, AI-driven research and patent intelligence for your innovation.

Sparse social network recommendation system sorting method based on virtual nodes

A virtual node and social network technology, applied in the field of sparse social network recommendation system ranking, can solve the problem that the graph neural network cannot fully play a role, and achieve the effect of low time complexity and improved accuracy

Pending Publication Date: 2022-01-11
TIANYI ELECTRONICS COMMERCE
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Graph neural network uses a message propagation mechanism, allowing adjacent nodes to pass messages through edges, which means that if the graph is sparse, that is, there are few edges in the graph, the graph neural network will not be able to fully function

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Sparse social network recommendation system sorting method based on virtual nodes
  • Sparse social network recommendation system sorting method based on virtual nodes
  • Sparse social network recommendation system sorting method based on virtual nodes

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0037] The present invention as Figure 1-3 As shown, the present invention provides a kind of virtual node-based sparse social network recommendation system sorting method, including the following process steps:

[0038] S1. Training sample generation:

[0039] 1). Based on the behaviors in the business system that can reflect the relationship between users, such as adding friends, following each other, sharing items, etc., generate social relationships between users;

[0040] 2). Collect the interactive operation behavior between users and items in the business system, including users and items exposed to users but not clicked, items exposed to users and clicked;

[0041] 3). Collect other contextual data, such as the time when the user clicks;

[0042] S2. Generate a social network graph according to the user's social relationship and user characteristics:

[0043] Take the user as the node V, and the social relationship between users as the edge E, generate an undirecte...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a virtual node-based sparse social network recommendation system sorting method, which is used for improving the click prediction accuracy of a recommendation system by introducing virtual nodes and an attention mechanism to assist message propagation on a sparse social network. The method has the advantages that 1, the virtual nodes are innovatively introduced as a bridge to assist message propagation on the sparse graph, a real graph structure can be restored to a certain extent, the user social relation graph can be fully utilized, the user characterizes and extracts richer information, and the accuracy of a recommendation system is improved; 2, the method of adding the virtual nodes can keep lower time complexity under the condition of solving graph sparsity, and compared with some complex technologies, the method has remarkable advantages in model training and model deployment.

Description

technical field [0001] The invention relates to the fields of graph neural network and Internet recommendation system, in particular to a sorting method for a virtual node-based sparse social network recommendation system. Background technique [0002] The Internet recommendation system has a wide range of applications in e-commerce, advertising, video and other fields. Through user characteristics, item characteristics, and the interaction between users and items, the recommendation system can judge the user's preference for items, so as to select the most suitable Items are recommended to users to improve item click rate and user satisfaction. [0003] Users may have similar preferences to their friends, or users may buy similar products based on their friends' recommendations. Therefore, the user's social relationship can greatly improve the traditional recommendation system that does not consider social interaction. Since user representation has an important impact on t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F16/9536G06N3/04G06Q30/06G06Q50/00
CPCG06F16/9536G06Q30/0631G06Q50/01G06N3/04
Inventor 周文彬傅剑文陈心童韩弘炀章建森
Owner TIANYI ELECTRONICS COMMERCE