Neural problem generation method for improving correlation

A Relevant, Question-Based Technique Applied to the Domain of Neural Question Generation for Improving Relevance

Active Publication Date: 2019-09-20
SUZHOU UNIV
View PDF6 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0009] The technical problem to be solved by the present invention is to provide a neural problem generation method that improves the correlation. Aiming at the problem of "generated problems have strong versatility and low correlation" in the QG model, we have proposed two improved methods: one One is to use a partial copy mechanism based on character overlap, and the words or their variant words appearing in the original text are prioritized through the degree of character overlap. The other is a reordering mechanism based on the QA model, and the quality of the generated questions is evaluated through the QA model and reorder based on that

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
  • Neural problem generation method for improving correlation
  • Neural problem generation method for improving correlation
  • Neural problem generation method for improving correlation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0030] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present invention and implement it, but the examples given are not intended to limit the present invention.

[0031] Background: QG model based on seq2seq

[0032] Such as figure 1 As shown, in the question generation system, the seq2seq framework is generally used to realize question generation, which consists of an encoder layer and a decoder, and an attention mechanism and a copy mechanism are added.

[0033] Embedding layer: We initialize each word in the training corpus as a word vector, which allows the model to handle it better. Word embeddings are usually pre-trained and their dimensions are predetermined. Taking the glove word vector we use as an example, each word can correspond to a 300-dimensional word vector.

[0034] Encoder layer: The encoder is usually a two-way LSTM, w...

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 neural problem generation method for improving correlation. The neural problem generation method for improving correlation comprises the steps: enabling a QG model based on seq2seq to be composed of an encoder layer and a decoder layer, and adding an attention mechanism and a copy mechanism; and providing a partial copy mechanism based on character overlapping. The method has the beneficial effects that through a partial copy mechanism based on the character overlap degree, the word level and the input document in the generated problem have higher overlap degree and correlation. Through a QA model-based reordering mechanism, higher scores can be given to the generated candidate questions with better quality, and the common questions difficult to answer can be filtered out.

Description

technical field [0001] The invention relates to the field of problem generation, in particular to a neural problem generation method for improving correlation. Background technique [0002] Question Generation (QG) is a very important problem in natural language processing. It is an important way to test whether the computer really understands the text, and it is widely used in various fields. QG can create a large number of QA pairs for Question Answering (QA), providing datasets for related tasks. At the same time, QG itself can also provide services for medical diagnosis systems, family education systems, etc. The input of the QG task usually consists of a document (or sentence) and an answer, and the output is to generate the most likely question given the document and the target answer. [0003] In general, the QG model is a sequence-to-sequence (seq2seq) structure consisting of an encoder (encoder) and a decoder (decoder). The encoder encodes the input document and t...

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): G06F16/332G06F17/27
CPCG06F16/3329G06F40/284
Inventor 熊德意邱嘉作
Owner SUZHOU UNIV
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