BiC2V-NFLAT: Chinese toponym recognition based on Bi-Char2Vec and non-flat-lattice transformer

Manzhen Yang

doi:10.1117/12.3034819

7 August 2024 BiC2V-NFLAT: Chinese toponym recognition based on Bi-Char2Vec and non-flat-lattice transformer

Manzhen Yang

Author Affiliations +

Proceedings Volume 13224, 4th International Conference on Internet of Things and Smart City (IoTSC 2024); 132241X (2024) https://doi.org/10.1117/12.3034819
Event: 4th International Conference on Internet of Things and Smart City, 2024, Hangzhou, China

Abstract

In response to the issues of missing context semantics and imbalanced datasets in Chinese toponym recognition tasks, this paper proposes a Chinese toponym recognition method based on Bi-Char2Vec and Non-Flat-Lattice Transformer. Initially, a new Chinese character embedding model, Bi-Char2Vec, is designed to capture the semantic representation of text in long sequences, mitigating the problem of missing context semantics in Chinese toponym recognition. Then, by utilizing Inter-Attention to interact between "character-word" attention, followed by encoding contextual information through the Self-Attention in Transformer; a global optimal tagging sequence is finally obtained by a Conditional Random Field layer. On public datasets, BIC2V-NFLAT improves the accuracy of Chinese toponym recognition, achieving an F1 score of 95%.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Manzhen Yang "BiC2V-NFLAT: Chinese toponym recognition based on Bi-Char2Vec and non-flat-lattice transformer", Proc. SPIE 13224, 4th International Conference on Internet of Things and Smart City (IoTSC 2024), 132241X (7 August 2024); https://doi.org/10.1117/12.3034819

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