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

Subjective question scoring model and scoring method based on deep learning BERT-CNN

A BERT--CNN, deep learning technology, applied in the subjective question scoring model and scoring based on deep learning BERT--CNN, subjective question scoring model and scoring field, can solve problems such as unreasonable and wasteful scoring structures, and achieve reduction Labor costs, improved performance, and the effects of overcoming inaccurate scoring results

Inactive Publication Date: 2019-10-08
KUNMING UNIV OF SCI & TECH
View PDF13 Cites 19 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the deficiencies in the prior art, one of the purposes of the present invention is to provide a subjective question scoring model based on deep learning BERT--CNN, through which the subjective questions are scored to solve the problem of manpower, Waste of financial resources and unreasonable scoring structure

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
  • Subjective question scoring model and scoring method based on deep learning BERT-CNN
  • Subjective question scoring model and scoring method based on deep learning BERT-CNN
  • Subjective question scoring model and scoring method based on deep learning BERT-CNN

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0026] Embodiment 1: This subjective question scoring model based on deep learning BERT--CNN includes BERT conversion word vector matrix module, CNN semantic feature extraction module, similarity calculation module, scoring module, wherein BERT conversion word vector matrix module, for Convert the answer text into a word vector matrix, then pass the word vector matrix of each answer to the CNN semantic feature extraction module, and the CNN semantic feature extraction module is used to obtain the semantic feature vector of the answer text, including convolutional layers and pooling layer to obtain the semantic feature vector of each answer text, and send it to the similarity calculation module to calculate the similarity value between the semantic feature vectors of each answer text, and then send the similarity value to the scoring module to determine the answer The score of the text.

[0027] Using this deep learning BERT--CNN subjective question scoring model to score, firs...

Embodiment 2

[0036] Embodiment 2: The method in this embodiment is the same as that in Embodiment 1, except that the pooling algorithms in step (1) and step (2) are both minimum pooling methods.

Embodiment 3

[0037] Embodiment 3: The method of this embodiment is the same as that of Embodiment 1, except that the pooling algorithms in step (1) and step (2) are both average pooling methods.

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 subjective question scoring model and scoring method based on deep learning BERT-CNN. The subjective question scoring model comprises a BERT conversion word vector matrix module, a CNN semantic feature extraction module, a similarity calculation module and a scoring module, wherein the BERT conversion word vector matrix module is used for converting an answer text into aword vector matrix, and then transmitting the word vector matrix of each answer to the CNN semantic feature extraction module; the CNN semantic feature extraction module is used for acquiring a semantic feature vector of the answer text, includes a convolution layer and a pooling layer, and is used for obtaining semantic feature vectors of the answer texts and then transmitting the signals to thesimilarity calculation module for calculating a similarity value between the answer text semantic feature vectors and transmitting the similarity value to the scoring module for determine the score ofthe answer text. The subjective question scoring model is applied to subjective question scoring, can effectively reduce the labor cost, and can solve the problems that due to the fact that only keyword matching is carried out in manual scoring, the scoring result is inaccurate, and scoring is not fair.

Description

technical field [0001] The invention relates to a scoring model and scoring method for subjective questions, in particular to a scoring model and scoring method for subjective questions based on deep learning BERT--CNN, belonging to the field of artificial intelligence. Background technique [0002] At present, subjective questions are scored only by matching keywords, that is, by extracting the keywords in the reference answers and the candidates’ answers, and then matching the keywords in the reference answers with the keywords in the candidates’ answers. If the matching rate is high , the score is high, otherwise the score is low or no score. [0003] Although this technology can extract keywords, it often ignores the meaning or semantics of words, and the relationship between words hides a lot of semantic information. Keyword-based matching will cause this part of information to be lost, resulting in inaccurate scoring results . Contents of the invention [0004] In ...

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): G06F17/27G06Q10/06G06Q50/20
CPCG06Q10/06393G06Q50/205G06F40/30
Inventor 侯开虎肖灵云戴洪涛杨少琦
Owner KUNMING UNIV OF SCI & TECH
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