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

Short text matching method and system based on global and local matching

A technology of local matching and matching methods, which is applied in the fields of instruments, biological neural network models, and electrical digital data processing, etc., and can solve problems such as resource waste, short text matching task characteristics are not exactly the same, and it is difficult to achieve good results.

Active Publication Date: 2020-09-15
SHENZHEN GRADUATE SCHOOL TSINGHUA UNIV
View PDF5 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In addition, the characteristics of different short text matching tasks are not exactly the same
For example, in the answer selection task of intelligent question answering, questions and answers are generally not synonymous or near-synonymous sentences. If the matching process is completely learned through a matching framework based on similarity, due to the data-driven characteristics of neural networks, in the amount of data It is difficult to have a good effect when the amount of data is insufficient, and it will often cause a waste of resources when the amount of data is sufficient

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
  • Short text matching method and system based on global and local matching
  • Short text matching method and system based on global and local matching
  • Short text matching method and system based on global and local matching

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0046] Embodiments of the present invention will be described in detail below. It should be emphasized that the following description is only exemplary and not intended to limit the scope of the invention and its application.

[0047] refer to figure 1 , the embodiment of the present invention provides a short text matching method based on global and local matching, comprising the following steps:

[0048] S1. Perform local matching, wherein the matching learning is performed on the internal matching model through the text;

[0049] S2. The local matching result is used as the attention weight of each text in the global matching to the node's own connection;

[0050] S3. Obtain the vector representation of the text pair based on the pre-trained language model, and use each matching pair as a node, and the vector of the matching pair as a feature of the graph node;

[0051] S4. Construct a text pair matching graph based on the text pair vector similarity, so that similar seque...

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 provides a short text matching method and system based on global and local matching, being characterized by constructing a matching graph, representing the whole short text pair as a graph node based on a pre-trained language model, carrying out cross-pair matching learning on the basis of graph attention network node classification, statically combining local matching and global matching through connection weights of graph attention network nodes, and dynamically combining the local matching and the global matching through joint training and iterative updating, so that local andglobal advantages are better exerted, and the short text matching method and system are suitable for different short text matching tasks.

Description

technical field [0001] The invention relates to computer applications, in particular to a short text matching method and system based on global and local matching. Background technique [0002] Short text matching is a fundamental problem in natural language processing. Many problems are essentially short text matching problems. Such as Paraphrase Identification, Natural Language Inference, Question Answering, etc. These problems are generally solved using a similarity matching framework. [0003] Traditional short text matching is mainly based on feature extraction and matching. In recent years, the application of deep learning has given full play to the advantages of representation learning. However, judging the degree of matching by the score of semantic recall, neural networks tend to assign high probabilities to short text pairs with similar patterns (words, phrases, structures, sentences, etc.). In addition, the characteristics of different short text matching task...

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): G06F40/279G06F40/30G06N3/04
CPCG06F40/279G06F40/30G06N3/045
Inventor 郑海涛徐子涵翟少鹏王栋周岚沈颖肖喜
Owner SHENZHEN GRADUATE SCHOOL TSINGHUA UNIV
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