Study on a simple summative attention mechanism used in BERT model for topic classification and sentiment analysis of car reviews

Jingja Huang; Xincheng Huang

doi:10.1117/12.3011061

2 November 2023 Study on a simple summative attention mechanism used in BERT model for topic classification and sentiment analysis of car reviews

Jingja Huang, Xincheng Huang

Author Affiliations +

Proceedings Volume 12919, International Conference on Electronic Materials and Information Engineering (EMIE 2023); 1291914 (2023) https://doi.org/10.1117/12.3011061
Event: 3rd International Conference on Electronic Materials and Information Engineering (EMIE 2023), 2023, Guangzhou,, China

Abstract

In the automobile industry, consumers' focus and satisfaction with automobiles are important reference directions for product optimization. This key information can be obtained by analyzing automobile evaluations. At present, end-to-end model training has problems such as large demand for data and long training time. This study proposes a summative attention mechanism to improve the Bert model, and transfers human prior knowledge to the model by summarizing vocabulary. The results show that the summative attention mechanism can make the model converge faster, significantly improve the classification accuracy of reviews, and has a good application value in the topic classification of car reviews.

(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Jingja Huang and Xincheng Huang "Study on a simple summative attention mechanism used in BERT model for topic classification and sentiment analysis of car reviews", Proc. SPIE 12919, International Conference on Electronic Materials and Information Engineering (EMIE 2023), 1291914 (2 November 2023); https://doi.org/10.1117/12.3011061

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available