Author name disambiguation method based on heterogeneous graph convolutional neural network embedding

A convolutional neural network, heterogeneous technology, applied in the field of big data, to achieve the effect of improving representativeness

Active Publication Date: 2019-11-29
COMP NETWORK INFORMATION CENT CHINESE ACADEMY OF SCI
View PDF5 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the technical difficulty to be solved in such a solution is how to learn high-quality publication representation vectors by using the various characteristics of publications and the relationship information between publications

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
  • Author name disambiguation method based on heterogeneous graph convolutional neural network embedding
  • Author name disambiguation method based on heterogeneous graph convolutional neural network embedding
  • Author name disambiguation method based on heterogeneous graph convolutional neural network embedding

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0018] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0019] The present invention adopts a network embedding method of an unsupervised heterogeneous graph convolutional neural network and a meta-path random walk strategy to disambiguate scholars' names. In the following embodiments, the name disambiguation publication benchmark database is selected as the publication database, and the present invention is further described in conjunction with the accompanying drawings. The method process of the present invention is as figure 1 shown.

[0020] Step 1: For an author name that needs to be disambiguated, collect all publications with the author name in the digital library, and construct a publication heterogeneous attribute network through the titles, author lists, and publications of these publications .

[0021] Treat each publication as a node in the heterogeneous attribute network. If there is a co-aut...

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 author name disambiguation method based on heterogeneous graph convolutional neural network embedding. The method comprises the following steps: 1) for a target author nameto be disambiguated, collecting publications made by the target author name, and then constructing a publication heterogeneous attribute network according to the collected title, author list and publication information of the publications; 2) according to the publication heterogeneous attribute network, generating a path accommodating publication node neighbor node text information through a random walk strategy based on a meta-path; 3) learning a representation vector of each publication based on a heterogeneous graph convolutional neural network embedding model according to the publication heterogeneous attribute network and the path; wherein the publication heterogeneous attribute network is used for representing publications of the target author name, 4) constructing a publication homogeneous network of the target author name according to the publication heterogeneous attribute network and the representation vector of the publications, and 5) dividing the publication homogeneous network to obtain a plurality of clusters, and the publications in the same cluster are publication sets of the same person.

Description

technical field [0001] The present invention relates to big data, knowledge map, entity disambiguation, graph neural network, heterogeneous network embedding technology field, specifically a network embedding method based on unsupervised heterogeneous graph convolutional neural network and meta-path random walk strategy Techniques for scholar name disambiguation. Background technique [0002] Nowadays, academic information mining in digital archives is becoming more and more important. When a user searches for an author's name in a digital library, what he wants to get is a search result that is both fast and accurate and that is relevant to the name. However, many search services in digital archives only retrieve a broad collection of publications, which leads to the problem of duplicate author names, that is, publications in this collection have the same author name, but these authors are not necessarily It's the same person. Using author name disambiguation technology ...

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): G06F16/9535G06N3/04G06N3/08
CPCG06F16/9535G06N3/08G06N3/045
Inventor 杜一乔子越周园春
Owner COMP NETWORK INFORMATION CENT CHINESE ACADEMY OF SCI
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