Unlock instant, AI-driven research and patent intelligence for your innovation.

Record question and answer classification method based on ERNIE and DPCNN

A classification method and recording technology, applied in text database clustering/classification, biological neural network models, instruments, etc., can solve problems such as increasing training complexity, not parallel processing, and not considering prior knowledge

Inactive Publication Date: 2020-10-23
HUNAN UNIV
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Most of the traditional text classification technologies use RNN or CNN models, but these models have great shortcomings. For example, the RNN model is a cyclic neural network, that is, the result of the next layer depends on the output of the previous layer. Generally speaking, a The word-by-word output is obviously not a friendly parallel processing method. At the same time, in recent years, some researchers have used ultra-deep CNN to classify text in a large-scale training data environment, which undoubtedly increases the complexity of training. Although the BERT model published in 2018 has achieved good results in the field of natural language, the BERT model is not friendly to the accuracy of Chinese classification and only predicts through context without considering the prior knowledge in Chinese sentences. Based on the above deficiencies, the present invention proposes a written question and answer classification method based on ERNIE and DPCNN

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
  • Record question and answer classification method based on ERNIE and DPCNN
  • Record question and answer classification method based on ERNIE and DPCNN
  • Record question and answer classification method based on ERNIE and DPCNN

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0028] Combine below figure 1 and figure 2 The concrete implementation process of the present invention is described, and content is as follows:

[0029] (1) Classification of written questions and answers

[0030] Such as figure 1 As shown, firstly classify the question and answer pairs in the transcript text, such as "personal situation", "criminal history", "physical condition" and other categories. The categories of different case transcripts are different. The categories of "drug source", "drug smoking method", etc., while the pickpocketing case records include "pickpocketing method", "pickpocketing location" and other categories;

[0031] (2) Data preprocessing

[0032] Since there may be situations in the original transcript information where both the above and the following are both question sentences or answer sentences, it is necessary to preprocess the original transcript text T, where T={T 1 , T 2 , T 3 ,...,T i ,...,T len(T)}, len(T) represents the numbe...

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 relates to a record question and answer classification method based on ERNIE and DPCNN. The method mainly comprises the following steps of: preprocessing a record text data set; inputting the processed data into an ERNIE model for training to obtain a word vector sequence; according to the method, inputting the obtained word vector sequence into the DPCNN model to be trained, and further extracting sentence features extracted by the ERNIE model by the DPCNN model. Therefore, the semantic representation of sentences is more accurate, and the accuracy of record question and answerclassification is improved.

Description

technical field [0001] The invention belongs to the technical field of natural language processing, and is a method for classifying question-answer pairs in transcripts based on ERNIE and DPCNN. Background technique [0002] With the passage of time, more and more transcripts have been accumulated. However, these transcripts contain a large amount of key case information. How to efficiently obtain important and valuable case information from these data has become a research hotspot. Questions and answers, search, information extraction, and criminal analysis are all applications of natural language processing. However, the classification of questions and answers in transcripts is the basis of these technologies, so the requirements for the accuracy of question and answer classification in transcripts are also very high. [0003] With the rapid development of the field of deep learning, the accuracy of text classification technology is constantly improving, such as spam class...

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): G06F16/35G06F16/332G06F16/33G06F40/279G06N3/04
CPCG06F16/35G06F16/3329G06F16/3346G06F40/279G06N3/045
Inventor 王莎莎彭鹏
Owner HUNAN UNIV