A Semantic Coding Method Based on Distracted Long Short-Term Memory Networks

A long-term and short-term memory, distraction technology, applied in semantic analysis, neural learning methods, biological neural network models, etc., can solve the problem of not establishing a link mechanism that integrates context information, and achieves improved accuracy and sentence correlation. , good integrity and fluency, and the effect of improving the degree of integrity

Active Publication Date: 2022-07-12
BEIJING INSTITUTE OF TECHNOLOGYGY +2
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although BiLSTM improves the problems of gradient explosion and gradient disappearance, the above two problems still exist when the sequence information is too long. At the same time, although BiLSTM can obtain the bidirectional feature information of the sequence, it only connects the bidirectional output without establishing a perfect Linking mechanism for fusing contextual information

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
  • A Semantic Coding Method Based on Distracted Long Short-Term Memory Networks
  • A Semantic Coding Method Based on Distracted Long Short-Term Memory Networks
  • A Semantic Coding Method Based on Distracted Long Short-Term Memory Networks

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0067] The present invention will be described in detail below according to the accompanying drawings and examples, but the specific embodiments of the present invention are not limited thereto.

[0068] This embodiment describes the process of applying the present invention "a semantic encoding method based on a long-term and short-term memory network with distraction" to a natural language generation and processing scenario.

[0069] The present invention trains and tests the model in a public dataset cMedQA and cMedQA1. cMedQA and cMedQA1 are a question and answer matching dataset for Chinese medical question and answer, which is widely used in some medical Chinese question and answer evaluation. The data of cMedQA comes from medical online forums, which include 54,000 questions and 100,000 corresponding answers. cMedQA1 is an extension of cMedQA, which contains 100,000 medical questions and about 200,000 corresponding answers.

[0070] The method provided in this embodim...

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 semantic coding method based on a distracted long-term and short-term memory network, which belongs to the field of natural language processing and generation. Aiming at the problems of semantic deviation, gradient disappearance, gradient explosion, and imperfect fusion context information existing in the prior art, the neural network used in the present invention adds a parameter sharing unit on the basis of BiLSTM, which enhances the model acquisition and fusion of bidirectional feature information. It adopts the activation function in the improved deep learning model to reduce the probability of the gradient problem; for the input and hidden layers, the model is constructed by using interactive space and extended LSTM, which enhances the ability of the model to integrate contextual information; A distraction mechanism of sentence structure information variables is introduced, which restricts the generation of semantics and improves the accuracy of semantics. The invention is suitable for the application of natural language generation such as automatic writing of news or headlines, robot customer service, conference or diagnosis report generation.

Description

technical field [0001] The invention relates to a semantic encoding method based on a distracted long-term and short-term memory network, and belongs to the field of natural language processing generation. Background technique [0002] Natural language generation is mainly used in the fields of human-computer dialogue, abstract, picture and video description, etc. It is the core technology for applications such as automatic news or headline writing, robot customer service, conference or diagnostic report generation. And the correctness of semantics is the key to generating language. [0003] Natural language generation mostly adopts the form of sequence conversion from encoding to decoding, which converts a sequence of information into another corresponding sequence of text. The process of hierarchical coding is divided into four steps: sentence semantic coding, text information coding, decoding, and sentence probability calculation. In the text generation step, sentence s...

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 & AuthorityPatents(China)
IPC IPC(8): G06F40/258G06F40/284G06F40/30G06N3/04G06N3/08
CPCG06F40/258G06F40/30G06F40/284G06N3/08G06N3/048G06N3/045
Inventor郭树理杨文涛韩丽娜王国威宋晓伟
OwnerBEIJING INSTITUTE OF TECHNOLOGYGY