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

Review Sentiment Analysis Method, Device and System Based on Knowledge Mutual Distillation

A sentiment analysis and sentiment technology, applied in the field of comment sentiment analysis based on mutual distillation of knowledge, which can solve the problems affecting the quality of comment sentiment analysis.

Active Publication Date: 2020-10-16
EAST CHINA JIAOTONG UNIVERSITY
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In view of the above situation, it is necessary to solve the problem that in the existing sentiment analysis model, because the three kinds of text representations are not integrated, the joint influence of the two attributes on the text emotion and the individual influence of each attribute are not used for comprehensive evaluation, resulting in the influence of Comments on issues with the quality of sentiment analysis

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
  • Review Sentiment Analysis Method, Device and System Based on Knowledge Mutual Distillation
  • Review Sentiment Analysis Method, Device and System Based on Knowledge Mutual Distillation
  • Review Sentiment Analysis Method, Device and System Based on Knowledge Mutual Distillation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0079] For this technical problem, the present invention proposes a comment sentiment analysis method based on mutual distillation of knowledge, please refer to Figure 2 to Figure 5 , The method of comment sentiment analysis based on mutual distillation of knowledge proposed in the first embodiment of the present invention.

[0080] What needs to be explained here is that the comment sentiment analysis method based on mutual distillation of knowledge proposed in the present invention is implemented based on the comment text representation learning neural network that integrates user and product attributes. Such as image 3 As shown, the hierarchical attention network (Hierarchical Attention Network, HAN) is used to learn the semantic representation of review text (document level); a simple multi-layer feed forward network (FFN) is used at the word level and sentence level. At the document level, user attribute information is integrated into the text representation; Attention Mec...

Embodiment 2

[0178] See Image 6 , The present invention also provides a comment sentiment analysis device based on mutual distillation of knowledge, wherein the device includes a first calculation module 111, a second calculation module 112, and an update convergence module 113 connected in sequence;

[0179] The first calculation module 111 is specifically used for:

[0180] It is used to determine a current training instance from the training data set, and calculate the first emotion prediction result based on the first classifier, wherein the first classifier takes the comment text with user attributes and product attributes as input, and the user The attributes represent different users corresponding to the comment text, and the product attributes represent different products pointed to by the review object in the review text;

[0181] The second calculation module 112 is specifically configured to:

[0182] In the case of ignoring the product attributes of the current training instance, the ...

Embodiment 3

[0186] See Figure 7 , The third embodiment of the present invention also provides a comment sentiment analysis system based on mutual distillation of knowledge, wherein the system includes a controller 11, the controller 11 and the first classifier 21, the second classifier 22, and The third classifier 23 is electrically connected, and there is bidirectional data transmission between each classifier and the controller 11. It is understandable that the controller 11 in this embodiment refers to the comment sentiment analysis device based on mutual distillation of knowledge in the second embodiment. In this embodiment, the controller 11 executes the comment sentiment analysis method based on mutual distillation of knowledge as described in the first embodiment.

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 proposes a comment sentiment analysis method, device and system based on mutual distillation of knowledge. The method includes: determining a current training instance from the training data set, and obtaining the first sentiment prediction result based on the first classifier; The classifier calculates the second emotion prediction result, and the third emotion prediction result is calculated based on the third classifier; according to the first, second, third emotion prediction results and the real category label, respectively calculate the overall corresponding to each of the three classifiers The cost function and each overall cost function are minimized to iteratively update the parameters of each classifier to achieve final convergence. The three classifiers in the present invention learn from each other and improve together based on the method of mutual knowledge distillation. At the same time, the common influence of the two attributes of the user and the product on the sentiment of the comment text and their individual influences are used to substantially improve the sentiment of comments on different occasions. Analyze performance.

Description

Technical field [0001] The invention relates to the technical field of computer intelligent analysis and processing, in particular to a method, device and system for comment sentiment analysis based on mutual distillation of knowledge. Background technique [0002] Document-level sentiment classification is a basic and important task in sentiment analysis and opinion mining. How to substantially improve the performance of sentiment analysis has become the main subject of research by experts and scholars. For example: a shopping website user "Zhang San" made a comment on the product "Huawei P30" he bought, "The photos taken by this phone are very clear", where "Zhang San" is the user attribute of this comment, and "Huawei P30" Is the product attribute of this review. Different users may have different vocabulary habits and their preference for different products may also be different. Therefore, it is necessary to consider the above two types of attribute information in the revi...

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 Patents(China)
IPC IPC(8): G06F16/35G06F40/205G06F40/126G06N3/04
CPCG06F16/355G06F40/205G06F40/126G06N3/049G06N3/045
Inventor 邬昌兴曹柳文
Owner EAST CHINA JIAOTONG UNIVERSITY
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