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

Method for automatically generating related work in proactive academic paper

An automatic generation and generative technology, applied in biological neural network models, natural language data processing, special data processing applications, etc., can solve problems such as poor cohesion and readability

Active Publication Date: 2018-11-06
BEIJING INSTITUTE OF TECHNOLOGYGY
View PDF7 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem of poor coherence and readability of related work generated by the extractive method, and proposes a method for automatically generating related work in generative academic papers

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
  • Method for automatically generating related work in proactive academic paper
  • Method for automatically generating related work in proactive academic paper
  • Method for automatically generating related work in proactive academic paper

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0128] The specific process of a generative automatic generation method for related work in academic papers is as follows: figure 1 As shown, the structural diagram of the HEDRA neural network involved in the flow chart is as follows figure 2 shown. This embodiment describes the flow of the method of the present invention and the structure and detailed parameters of the neural network involved.

[0129] A generative automatic generation method for related work in academic papers used in this embodiment specifically includes three sequentially executed processes of data collection phase, training phase and testing phase. The flow chart is as follows figure 1 Shown:

[0130] Among them, the data collection stage is to construct large-scale corpus pairs for neural network training and testing;

[0131] In the training phase, a hierarchical encoder-decoder based on residual attention is constructed, that is, HierarchicalEncoder-Decoder based on Residual Attention, abbreviated ...

Embodiment 2

[0210] On the test set, the method of the present invention marks each generated sentence with a number of cited references, while the traditional extraction method can only assign a reference number to each generated sentence. The present invention has counted the average accuracy rate, average recall rate and average F value of reference numbers in all sentences generated in related work, and compared with the traditional extraction method, and the specific results are as shown in table 3:

[0211] Table 3 uses the reference number generated by the method proposed by the present invention to compare with the number generated by the extractive method

[0212]

[0213] The experimental result of table 3 shows, the method that the present invention adopts when assigning the reference number to each sentence in the related work of generation, precision rate, recall rate and F value all have very big promotion compared with traditional extraction method, Especially in terms of...

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 present invention discloses a method for automatically generating the related work in a proactive academic paper, and belongs to the technical field of computer natural language processing. The specific operation steps comprise: 1 constructing a large-scale corpus based on the related work in the English academic papers and the abstracts of the references cited in the related work; 2 trainingthe neural network of a hierarchical encoder-decoder structure based on the residual attention according to the constructed corpus, and performing multitask learning according to the loss value of thereference number cited in the sentence and the loss value of the word in the sentence; and 3 inputting multiple abstracts into the trained neural network, and generating the corresponding related work word by word and sentence by sentence. Compared with the conventional extraction method, by using the method for automatically generating the related work in a proactive academic paper proposed by the present invention, not only each word in the related work can be automatically generated, but also a plurality of reference numbers can be assigned to each sentence in the generated related work.

Description

technical field [0001] The invention relates to a method for automatically generating related work in generative academic papers, in particular to a method for automatically generating Related Work in generative English academic papers, and belongs to the technical field of computer natural language processing. Background technique [0002] Related work is a chapter in the writing of academic papers, which refers to the author's summary and generalization of previous work in the field covered by the paper. Generally speaking, related works are short in length, but contain the main ideas of dozens of papers in this field. For a scientific researcher, reading the related work of an academic paper can grasp the research status of related fields in a short period of time and better grasp the direction of scientific research. Therefore, relevant work in academic papers plays a very important role in outlining research methods and guiding readers to quickly understand the researc...

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): G06F17/27G06F17/30G06N3/04
CPCG06F40/211G06F40/284G06N3/045
Inventor 毛先领姜晓健冯博思魏骁驰
Owner BEIJING INSTITUTE OF TECHNOLOGYGY
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