Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Internet food entity alignment method and system based on graph neural network

A technology of neural network and entity pairing, which is applied in the field of knowledge graph and neural network, can solve problems such as the impact of alignment effect, ignore graph structural heterogeneity, and influence effect, so as to improve quality, improve utilization efficiency and entity alignment accuracy, The effect of reducing the influence of heterogeneity

Active Publication Date: 2021-09-03
BEIJING TECHNOLOGY AND BUSINESS UNIVERSITY
View PDF3 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, different knowledge graphs have different graph structures due to their different emphases. The above methods ignore the heterogeneity of the graph structure; at the same time, there are entity attribute information in the knowledge graph, and this part of information is not used when entities are aligned.
[0005] To sum up, the current embedding-based entity alignment method relies heavily on alignment seeds, but in reality the number of entity-pair seeds is limited, while ignoring the heterogeneity of graph structures in different knowledge graphs; in addition, attribute information and relationship information It will also affect the entity alignment effect, and the attribute information of the entity is not used when learning the entity embedding
These problems will affect the effect of entity alignment

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
  • Internet food entity alignment method and system based on graph neural network
  • Internet food entity alignment method and system based on graph neural network
  • Internet food entity alignment method and system based on graph neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0026] Such as figure 1 As shown, a method for aligning Internet food entities based on a graph neural network provided by an embodiment of the present invention includes the following steps:

[0027] Step S1: Obtain food information through a web crawler, extract triples among them, construct two knowledge graphs KG1 and KG2 to be aligned, and separate the triples in KG1 and KG2 to obtain relational triples and attribute triplets;

[0028] Step S2: Deduce the rules for the relational triples, transfer the rules between KG1 and KG2, construct a new relational triplet, update the relational triplet, and obtain the updated relational triplet;

[0029] Step S3: Calculate and obtain two adjacency matrices with self-attention mechanism weights and cross-attention mechanism weights according to the updated relation triplet, as well as self-attention mechanism and cross-attention mechanism; and pre-train word vectors according to bert , query the vector table to obtain the entity wo...

Embodiment 2

[0099] Such as Figure 8 As shown, the embodiment of the present invention provides a graph neural network-based Internet food entity alignment system, including the following modules:

[0100] The separation triplet module is used to obtain food information through web crawlers, extract the triplets, construct two knowledge graphs KG1 and KG2 to be aligned, and separate the triplets in KG1 and KG2 to obtain relational triplets and attribute triplet;

[0101] Updating the relational triplet module, which is used to reason the relational triplet to obtain rules, transfer the rules between KG1 and KG2, construct a new relational triplet, and obtain an updated relational triplet;

[0102] Obtain the entity structure feature vector and relation feature vector module, which is used to calculate two weights with self-attention mechanism weight and cross-attention mechanism weight according to the updated relation triplet, as well as self-attention mechanism and cross-attention mech...

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 relates to an internet food entity alignment method and system based on a graph neural network, and the method comprises the steps: S1, separating triads in two knowledge maps KG1 and KG2 to be aligned, and obtaining a relation triad and an attribute triad; s2, updating a relation triple; s3, obtaining an entity structure feature vector and a relation feature vector; s4, obtaining an entity feature vector; and S5, constructing an entity relationship alignment model, calculating the distance between entity feature vectors to realize entity alignment, and calculating the distance between relationship feature vectors to realize relationship alignment. According to the method provided by the invention, the graph structure of the knowledge graph is complemented and trimmed, so that the influence of the isomerism of graphs among the knowledge graphs is reduced, and meanwhile, the utilization efficiency of entity seeds and the accuracy of entity alignment are improved under the condition that the entity seeds are limited.

Description

technical field [0001] The invention relates to the fields of knowledge graphs and neural networks, in particular to a method and system for aligning Internet food entities based on graph neural networks. Background technique [0002] Knowledge graphs are the basis of natural language processing tasks such as machine reading, machine translation, and recommendation systems. Different research fields establish different knowledge graphs. The purpose and focus of knowledge graphs are different when creating knowledge graphs. As a result, different knowledge graphs will contain a lot of complementary information. Integrating complementary information in these knowledge graphs will improve the efficiency of knowledge utilization, but the same entity may have different representations in different knowledge graphs, and entity alignment is needed to solve this problem. [0003] In recent years, embedding-based methods are mainly used to align entities in different knowledge graphs...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/22G06F16/28G06F16/951G06N3/04G06N3/08
CPCG06F16/2237G06F16/288G06F16/951G06N3/082G06N3/048Y02P90/30
Inventor 左敏薛明慧张青川颜文婧
Owner BEIJING TECHNOLOGY AND BUSINESS UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com