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

Travel comment opinion mining method based on BERT

A technology of opinion mining and opinion words, which is applied in the field of BERT-based tourism review opinion mining, can solve problems such as unclear description goals and missing effective evaluation information, and achieve the effect of improving expression ability

Pending Publication Date: 2021-04-02
UNIV OF ELECTRONIC SCI & TECH OF CHINA
View PDF8 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

By extracting opinion words instead of aspect words, the problem of missing effective evaluation information caused by the lack of aspect words is made up for, and the problem of unclear description goals caused by the lack of aspect words is made up for by category classification, and applied in actual review 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
  • Travel comment opinion mining method based on BERT
  • Travel comment opinion mining method based on BERT
  • Travel comment opinion mining method based on BERT

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0041] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

[0042] Such as figure 1 Shown, a kind of BERT-based tourism comment mining method of the present invention comprises the following steps:

[0043] S1, process the input comment text, and convert the text into a token sequence that meets the conditions; such as figure 2 As shown, it specifically includes the following sub-steps:

[0044] S11. Load the vocab provided in the selected BERT pre-training model, convert the comment text into the form of a digital token, and complete preliminary tokenization.

[0045] Text vectorization is the basis for NLP to process text data. In the process of using the previous pre-training model, in order to query the efficiency of the corresponding vector and balance the training accuracy, a statistics is usually made on the vocabulary used in the training data, and the occurrence Words whose fre...

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 travel comment opinion mining method based on BERT. The method comprises the following steps of S1, processing an inputted comment text, and converting the text into a tokensequence meeting conditions, S2, performing hierarchical calculation processing on the input sequence by applying BERT to obtain encoded context representation, S3, inputting the obtained context representation into a pointer network for calculation to obtain an opinion word starting position candidate set and an opinion word ending position candidate set, S4, pairing the candidate set according to the classification result and the relative distance to obtain a final opinion word position, and S5, combining the opinion words and the corresponding classification results together to obtain a complete 'class, opinion word' opinion expression. According to the method, the problems of opinion loss and incompleteness caused by aspect word missing in a traditional fine-grained opinion mining method are solved in a mode of directly extracting opinion words and carrying out class labeling, and the method is applied to online tourism comments.

Description

technical field [0001] The invention relates to a BERT-based mining method for travel comments. Background technique [0002] The website has a huge amount of tourist reviews, which can be mined with the help of natural language processing technology. Aspect-Based Sentiment Analysis (ABSA) is an effective method for fine-grained opinion mining. ABSA aims to determine the opinions (including opinion words and sentiment polarity) of reviews on specific aspects (including aspect words and descriptive categories). For example, in the comment "The scenery on the Golden Summit of Mount Emei is good, but the price is really expensive", "scenery" is an aspect word, and the category described is scenery, and "nice" is an opinion word describing the aspect word, and the emotional polarity involved is positive. Towards. [0003] In 2014, SemEval introduced aspect-level sentiment analysis as a comprehensive evaluation task, which has provided a general evaluation framework for Englis...

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): G06F16/35G06F40/126G06F40/242G06F40/284G06Q50/14
CPCG06F16/35G06F40/126G06F40/242G06F40/284G06Q50/14
Inventor 江维蔡玉舒詹瑾瑜周星志温翔宇宋子微孙若旭范翥峰廖炘可
Owner UNIV OF ELECTRONIC SCI & TECH OF CHINA
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