Header column entity relationship matching method based on deep learning multi-head selection model

A technology of model selection and deep learning, applied in the direction of neural learning methods, biological neural network models, and other database retrievals, can solve problems such as ignoring the semantics of table headers and high startup costs of the scheme, and achieve the effect of low startup costs

Active Publication Date: 2021-09-10
南京烽火星空通信发展有限公司
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0002] The header column entity relationship matching technology is to determine the corresponding relationship between the two columns of the table. This technology plays an important role in table information mining. The existing technical solution is to use data item collision for judgment, ignoring the semantic information of the table header. This method needs to maintain a benchmark database by itself before data collision can be performed. The startup cost of the scheme is relatively high, and the header semantics are not used for auxiliary judgment

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
  • Header column entity relationship matching method based on deep learning multi-head selection model
  • Header column entity relationship matching method based on deep learning multi-head selection model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0027] Example: such as Figure 1-2 As shown, the present invention is based on the table header column entity relationship matching method of the deep learning multi-head selection model, including several steps:

[0028] Step 1: Define data entity attribute categories for the data items in the table, such as time, name, company name, etc., and build a regular recognition method;

[0029] Step 2: Construct the artificial features of any combination of two columns in the table header. The construction method of the artificial features can be selected according to the needs of the actual scene;

[0030] Step 3: After passing the header character sequence and the data attribute sequence corresponding to the header through the respective embedding layers, the merged vector is used as the input of the next encoding layer;

[0031] Step 4: The coding layer adopts the bi-lstm model structure, and the output is the code of each position of the header sequence fused with the context ...

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 header column entity relationship matching method based on a deep learning multi-head selection model, which comprises the following steps of: defining data entity attribute categories such as time, names, company names and the like for data items of a table, and constructing a regular recognition method; constructing artificial features of any two column combinations of the header; enabling the header character sequence and the data attribute sequence corresponding to the header to pass through respective embedding layers; enabling the coding layer to adopt a bi-lstm model structure; carrying out pairwise combination splicing on the context coding information at any position based on a multi-head selection mechanism; calculating a binary loss value of any two positions of the header sequence for each relation category; and reserving and serving the model with the loss value converged to the best as a model used for prediction. The invention discloses a header column entity relationship matching method based on a deep learning multi-head selection model. The method has certain convenience through a header and a model.

Description

technical field [0001] The present invention relates to the technical field of header column entities, in particular to a header column entity relationship matching method based on a deep learning multi-head selection model. Background technique [0002] The header column entity relationship matching technology is to determine the corresponding relationship between the two columns of the table. This technology plays an important role in table information mining. The existing technical solution is to use data item collision for judgment, ignoring the semantic information of the table header. This method needs to maintain a benchmark database by itself before data collision can be performed. The startup cost of the scheme is relatively high, and the header semantics are not used for auxiliary judgment. Therefore, we improve this and propose a method for matching the entity relationship of header columns based on the deep learning multi-head selection model. Contents of the i...

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): G06F16/903G06F16/906G06F40/126G06F40/18G06N3/08
CPCG06F16/90344G06F16/906G06F40/126G06F40/18G06N3/084
Inventor 高永伟李曙光宋万军姜广栋杨万刚李峰蔡晨陈玉冰皮乾东黄昌彬杜俊杰张鑫涛
Owner 南京烽火星空通信发展有限公司
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