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

Entity and relation joint extraction method

A technology of relation and entity, applied in the field of joint extraction of entity and relation

Active Publication Date: 2021-08-20
GUANGDONG UNIV OF TECH
View PDF10 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The embodiment of the present application provides a method for joint extraction of entities and relationships, which allows the relationship between two sub-tasks to be considered, thereby improving the accuracy and recall of relationship extraction, and solving the problem that one type of entity involves multiple types of entity relationships

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 and relation joint extraction method
  • Entity and relation joint extraction method
  • Entity and relation joint extraction method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0050] Therefore, this application proposes a new end-to-end method-a method of joint entity and relationship extraction combined with fine-tuning Bert model and graph convolutional neural network. This application uses single-layer Bi-LSTM and stacked Bi-LSTM to obtain text context features and deep context features respectively, and Bi-GCN obtains text context dependency information for entity naming recognition, and maps entity naming recognition results to label embedding Concatenation with deep contextual features for relation extraction. The relationship prediction result is used as the adjacency matrix of Bi-GCN and the output of the single-layer Bi-LSTM is updated to obtain context features, and then the second stage of entity recognition and relationship extraction is performed as the final result. This application fully considers the relationship between two subtasks, thereby improving the accuracy and recall rate of relationship extraction, and solving the problem t...

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 an entity and relation joint extraction method, which comprises the following steps of: converting a text into a low-dimensional dense vector to obtain a second text; extracting context features; analyzing the dependency relationship and establishing a dependency tree; taking the dependency tree as an adjacent matrix, extracting a first node feature of a graph of the dependency tree, converting the first node feature into a tag sequence, searching the tag sequence, and predicting a first-stage named entity result; extracting deep text features of the second text; splicing the named entity result and the deep text feature, and predicting the relationship; converting a named entity result and a relation prediction result into a graph structure; taking the graph structure as an adjacent matrix, and extracting a second node feature of the adjacent matrix; converting the updated text feature into a tag sequence, searching the tag sequence, and predicting a named entity result of a second stage of the text feature; and embedding the feature into the deep text feature, and predicting the relationship of the embedding results. According to the method, the relationship between the sub-tasks is considered, and the problem that one entity relates to a multi-entity relationship is solved.

Description

technical field [0001] The present application relates to the technical field of natural language processing, in particular to a method for joint extraction of entities and relationships. Background technique [0002] As a comprehensive management method, supply chain management is a research hotspot in large enterprises. The current domestic supply chain management methods mainly rely on manual management, but often due to too many tasks and too heavy management knowledge, the efficiency of supply chain management is low. Since a large amount of supply chain management knowledge exists in unstructured data, such as news, literature and so on. Comprehensively use deep learning, NLP and other artificial intelligence technologies to process supply chain-oriented big data, especially unstructured data, and realize the automatic construction technology of large-scale semantic knowledge base of supply chain. The key technology for building a semantic knowledge base is how to ex...

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): G06F40/295G06N3/04
CPCG06F40/295G06N3/044
Inventor 程良伦林锐明王涛王卓薇邓健峰周佳乐
Owner GUANGDONG UNIV OF TECH