LE QUY DON
Technical University
VietnameseClear Cookie - decide language by browser settings

Vietnamese Document Classification Using Hierarchical Attention Networks

Nguyen, K.D.T. and Viet, A.P. and Hoang, T.H. (2020) Vietnamese Document Classification Using Hierarchical Attention Networks. In: 7th International Conference on Frontiers of Intelligent Computing: Theory and Applications, FICTA 2018, 29 November 2018 through 30 November 2018.

Text
Vietnamese Document Classification Using Hierarchical Attention Networks..pdf

Download (518kB) | Preview

Abstract

Automatic document classification is considered to be an important part of managing and processing document in digital form, which is increasing. While there are a number of studies addressing the problem of English document classification, there are few studies that deal with the problem of Vietnamese document classification. In this paper, we propose to employ a hierarchical attention networks (HAN) for Vietnamese document classification. The HAN network has the two-level architecture with attention mechanisms applied to the word level and sentence level from which it reflects the hierarchical structure of the document. Experimental results are conducted on the Vietnamese news Database which is collected from the Vietnamese news Web sites. The results show that our proposed method is promising in the Vietnamese document classification problem. © Springer Nature Singapore Pte Ltd. 2020.

Item Type: Conference or Workshop Item (Paper)
Divisions: Faculties > Faculty of Information Technology
Identification Number: 10.1007/978-981-13-9920-6_13
Uncontrolled Keywords: Computation theory; Intelligent computing; Attention mechanisms; Digital forms; Document Classification; Hierarchical structures; News web sites; Sentence level; Vietnamese; Word level; Information retrieval systems
Additional Information: Conference code: 232789. Language of original document: English.
URI: http://eprints.lqdtu.edu.vn/id/eprint/9183

Actions (login required)

View Item
View Item