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

Relation extraction method suitable for small samples

A relationship extraction, small sample technology, applied in the field of knowledge graph construction, can solve problems such as inescapability, avoid time-consuming, money-consuming, and reduce manual labeling of data.

Pending Publication Date: 2020-05-08
南京中新赛克科技有限责任公司
View PDF2 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

But for a variety of relationship types, in dealing with the relationship extraction problem in a specific field, even if the Bert model can be used to learn basic language knowledge and improve the extraction results, it still cannot get rid of the data that needs to rely on marking

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
  • Relation extraction method suitable for small samples
  • Relation extraction method suitable for small samples
  • Relation extraction method suitable for small samples

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0028] A method of relation extraction adapted to small samples, comprising the following steps:

[0029] (1) Obtain training data;

[0030] (2) General Domain Relational Knowledge Model Training;

[0031] (3) Domain-specific relationship extraction model training.

[0032] In step (1), the training data is acquired specifically as follows: the training data comes from two parts, one is public relational labeling data, and the other is training data generated based on weak supervision;

[0033] (11) Use crawler tools to collect non-formatted text data on the Internet;

[0034] (12) Obtain triplet data (including relationship name and entity pair) on the public dataset Freebase;

[0035] (13) Obtain entity pair and its corresponding sentence by the named entity recognition method of NLP by text data;

[0036] (14) Using the remote supervision method, the entity pair and its corresponding sentence are assigned a relationship, and the same entity pair and its sentence are pla...

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 relation extraction method suitable for small samples. The relation extraction method comprises the following steps: (1) obtaining training data; (2) training a general domain relation knowledge model; and (3) training a specific domain relation extraction model. Common knowledge contained in various relationships is obtained by utilizing a general domain relationship knowledge module, and samples are automatically generated based on remote supervision by utilizing an open-source knowledge graph and combining with unsupervised noise reduction data to train relationship knowledge models of general and specific domains; a general domain relation knowledge module is adopted to learn general knowledge contained in various relations; training samples are automaticallygenerated on the basis of remote supervision, and noise data is reduced in combination with unsupervised data, so that manual annotation data is reduced; when the relation knowledge model is generated, a large amount of manual marking data does not need to be obtained, time and money consumption caused by a large amount of manual marking is avoided, and a relation extraction task in a specific field can be completed through a small amount of marking data in the specific field.

Description

technical field [0001] The present invention relates to the technical field of knowledge map construction, in particular to a method for extracting relationships adapted to small samples. Background technique [0002] Information extraction is an important part of natural language processing, especially in today's information society, it is particularly meaningful to extract useful information from massive data. Information extraction can be divided into entity extraction, relationship extraction and event extraction. The relationship extraction is based on the extracted entity pairs to determine whether there is a certain relationship between the entity pairs, and no relationship is also regarded as a special relationship. [0003] As relational extraction shifts from limited relational types to various relational types in the open domain, data sources shift from standard corpora to massive network data. Traditional pattern-matching-based methods cannot adapt to multiple ...

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): G06F16/36G06F16/35G06F16/951G06F40/295G06F40/211G06K9/62G06N3/04G06N3/08
CPCG06F16/367G06F16/35G06F16/951G06N3/088G06N3/045G06F18/23
Inventor 卓可秋杨秀燕
Owner 南京中新赛克科技有限责任公司
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