Entity relation joint extraction method and device, computer terminal and storage medium

A technology of entity relationship and relationship, applied in computing, instruments, biological neural network models, etc., can solve problems such as large amount of calculation, insufficient integration, and improvement of triplet extraction accuracy, so as to improve accuracy, improve features, Avoiding the effects of mispassing and redundant entities

Active Publication Date: 2022-04-12
CHENGDU SHULIANYUNSUAN TECH CORP
View PDF17 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, using a one-dimensional vector to represent the feature information of each text requires a huge amount of calculation and cannot fully integrate the features of the two sub-tasks of entity recognition and relationship extraction, so that the accuracy of triplet extraction is limited.

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
  • Entity relation joint extraction method and device, computer terminal and storage medium
  • Entity relation joint extraction method and device, computer terminal and storage medium
  • Entity relation joint extraction method and device, computer terminal and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0057] figure 1 It is a schematic flow chart of the entity-relationship joint extraction method in this embodiment, and the method includes the following steps:

[0058] In step S100, a text tensor is obtained based on the text data.

[0059] In this embodiment, take the text data of "Zhang San was born in Chengdu." A feature vector representing the character to be processed can be obtained, and the dimension of the vector is different according to the number of extracted features.

[0060] After entering the text feature model, you can get the text tensor. For example, if you use bert-base-chinese to extract the input text features, you can extract a 768-dimensional feature vector. The above 9 characters (including periods) are spliced ​​​​in order. These vectors, then Is a 9*768 tensor matrix, that is, a text tensor.

[0061] Step S200, according to the text tensor, obtain the head feature tensor and tail feature tensor of the text;

[0062] Copy two copies of the text t...

Embodiment 2

[0092] The present application also provides an entity-relationship joint extraction device, such as Figure 5 shown, including:

[0093] Pre-extraction module 10, for obtaining text tensor based on text data;

[0094] The segmentation module 20 is used to obtain the head feature tensor and the tail feature tensor of the text according to the text tensor;

[0095] A fusion module 30, configured to perform feature fusion on the head feature tensor and the transposed tail feature tensor to obtain a fusion tensor;

[0096] A scoring module 40, configured to input the fusion tensor into the convolutional neural network to obtain a scoring tensor;

[0097] The extraction module 50 is configured to input the scoring tensor into the prediction model, obtain the probability distribution data of each element in the scoring tensor in the entity and relation label space, and output the extraction result according to the probability distribution data.

[0098] The present application a...

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 embodiment of the invention discloses an entity relation joint extraction method and device, a computer terminal and a storage medium. The entity relationship joint extraction method comprises the following steps: obtaining a text tensor based on text data; acquiring a head feature tensor and a tail feature tensor of the text according to the text tensor; performing feature fusion on the head feature tensor and the transposed tail feature tensor to obtain a fusion tensor; inputting the fusion tensor into a convolutional neural network to obtain a scoring tensor; and inputting the scoring tensor into a prediction model to obtain probability distribution data of each element in the scoring tensor in the tag space of the entity and the relationship, and outputting an extraction result according to the probability distribution data. In the natural language processing, the convolutional neural network is used on the two-dimensional matrix to further extract features, so that subject, object and relation joint extraction is realized, the calculation amount can be greatly reduced, and the accuracy of triple extraction can be improved.

Description

technical field [0001] The invention relates to the field of natural language processing, in particular to a method, a device, a computer terminal and a storage medium for joint extraction of entity relations. Background technique [0002] There are two most commonly used entity and relationship extraction methods, one is the pipeline extraction method represented by pipeline, and the other is the entity-relationship joint extraction method. The pipeline extraction method considers that entities and relations belong to different semantic spaces. The extraction of entities and relations needs to be divided into two tasks, namely named entity recognition and relation classification. This extraction method has problems such as error transmission and redundant entities. The entity-relationship joint extraction method considers that entities and relations belong to the same semantic space, and a model needs to be jointly trained after fully integrating entity features and relatio...

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): G06F40/279G06N3/04
Inventor 不公告发明人
Owner CHENGDU SHULIANYUNSUAN TECH CORP
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap