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

Document-level relation extraction method based on graph neural network and reasoning path

A neural network and relationship extraction technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve the problem of affecting model performance, ignoring the solution node reasoning relationship, and considering two node reasoning paths that are not displayed in the graph neural network and other issues to achieve the effect of improving performance

Pending Publication Date: 2022-07-29
HARBIN INST OF TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

2) The computational cost of the model
When two nodes are far apart, it will seriously affect the performance of the model;
[0007] 2. The graph neural network does not show the reasoning path between two nodes, ignoring the reasoning relationship between nodes

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
  • Document-level relation extraction method based on graph neural network and reasoning path
  • Document-level relation extraction method based on graph neural network and reasoning path
  • Document-level relation extraction method based on graph neural network and reasoning path

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0030] combine figure 1 and figure 2 , the present invention proposes a document-level relationship extraction method based on a graph neural network and an inference path, the method comprising:

[0031] Step 1. Convert an input document into a graph structure based on heuristic rules;

[0032] Step 2, using the path search algorithm to extract multiple paths between different entity pairs in the graph structure;

[0033] Step 3, encoding the input document by using the neural network encoder, and obtaining the vector representation of the nodes in the graph, and using the graph neural network to update the vector representation of the nodes in the graph;

[0034] Step 4: Obtain the vector representation of path information between entity pairs in the graph structure;

[0035] Step 5: Judge the relationship between the entity pairs, and use the labeled data to train the deep learning model.

[0036] The graph structure converted in step 1 is a heterogeneous graph structu...

Embodiment 2

[0068] In this embodiment, steps 1, 2, 3, and 5 are the same as those in the first embodiment. In step 4 of this embodiment, a vector representation of path information between entity pairs in the graph structure is obtained. In addition to using the nodes output by the graph neural network to represent the features of the nodes in the path, consider directly using the initial representation of the nodes as the features of the nodes in the path (such as figure 2 left).

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 provides a document-level relation extraction method based on a graph neural network and a reasoning path. The method aims at solving the problem that a common graph model method in document-level relation extraction can only pay attention to the characteristics of entity local features and cannot well represent global features between two entities. The method comprises the following specific steps of: 1, converting an input document into a graph structure based on a heuristic rule; step 2, extracting a plurality of paths between different entity pairs from the construction graph structure by using a path search algorithm; 3, encoding the input document by using a neural network encoder, obtaining vector representations of nodes in the graph, and updating the vector representations of the nodes in the graph by using a graph neural network; step 4, obtaining path information vector representation between entity pairs in the graph structure; and 5, judging the relationship between the entity pairs, and training a deep learning model by using marked data. The invention belongs to the field of natural language processing.

Description

technical field [0001] The invention belongs to the technical field of natural language processing, and in particular relates to a document-level relationship extraction method based on a graph neural network and an inference path. Background technique [0002] The task of document-level relationship extraction is to determine whether there are some predefined relationships among all entity pairs given a paragraph containing multiple sentences, entities appearing in the paragraph, and entities appearing in the paragraph. Compared with sentence-level relationship extraction, document-level relationship extraction has the following technical difficulties: 1) Document-level relationship requires a variety of different reasoning methods, including intra-sentence relationship extraction, referential reasoning, logical reasoning, and common sense reasoning. How to design a better document-level relation extraction model, effectively synthesize useful information from longer contex...

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): G06F40/211G06F40/295G06N3/04G06N3/08
CPCG06F40/211G06F40/295G06N3/08G06N3/044
Inventor 赵铁军陈科海徐旺曹海龙朱聪慧徐冰杨沐昀
Owner HARBIN INST OF TECH