BIT-WOW at NLPCC-2022 Task5 Track1: Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network

Bo Wang; Yi Fan Lu; Xiaochi Wei; Xiao Liu; Ge Shi; Changsen Yuan; Heyan huang; Chong Feng; Xianling Mao

doi:10.1007/978-3-031-17189-5_16

BIT-WOW at NLPCC-2022 Task5 Track1: Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network

Bo Wang, Yi Fan Lu, Xiaochi Wei, Xiao Liu, Ge Shi, Changsen Yuan, Heyan huang^*, Chong Feng, Xianling Mao

^*Corresponding author for this work

School of Computer Science and Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

This paper describes the system proposed by the BIT-WOW team for NLPCC2022 shared task in Task5 Track1. The track is about multi-label towards abstracts of academic papers in scientific domain, which includes hierarchical dependencies among 1,530 labels. In order to distinguish semantic information among hierarchical label structures, we propose the Label-aware Graph Convolutional Network (LaGCN), which uses Graph Convolutional Network to capture the label association through context-based label embedding. Besides, curriculum learning is applied for domain adaptation and to mitigate the impact of a large number of categories. The experiments show that: 1) LaGCN effectively models the category information and makes a considerable improvement in dealing with a large number of categories; 2) Curriculum learning is beneficial for a single model in the complex task. Our best results were obtained by an ensemble model. According to the official results, our approach proved the best in this track.

Original language	English
Title of host publication	Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings
Editors	Wei Lu, Shujian Huang, Yu Hong, Xiabing Zhou
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	192-203
Number of pages	12
ISBN (Print)	9783031171888
DOIs	https://doi.org/10.1007/978-3-031-17189-5_16
Publication status	Published - 2022
Event	11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022 - Guilin, China Duration: 24 Sept 2022 → 25 Sept 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13552 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022
Country/Territory	China
City	Guilin
Period	24/09/22 → 25/09/22

Keywords

Curriculum learning
Graph convolutional network
Hierarchical multi-label classification
Label embedding

Access to Document

10.1007/978-3-031-17189-5_16

Cite this

Wang, B., Lu, Y. F., Wei, X., Liu, X., Shi, G., Yuan, C., huang, H., Feng, C., & Mao, X. (2022). BIT-WOW at NLPCC-2022 Task5 Track1: Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network. In W. Lu, S. Huang, Y. Hong, & X. Zhou (Eds.), Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings (pp. 192-203). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13552 LNAI). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-17189-5_16

Wang, Bo ; Lu, Yi Fan ; Wei, Xiaochi et al. / BIT-WOW at NLPCC-2022 Task5 Track1 : Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network. Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. editor / Wei Lu ; Shujian Huang ; Yu Hong ; Xiabing Zhou. Springer Science and Business Media Deutschland GmbH, 2022. pp. 192-203 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{35274c379f3c4f1ebb26f6cc3784b9a0,

title = "BIT-WOW at NLPCC-2022 Task5 Track1: Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network",

abstract = "This paper describes the system proposed by the BIT-WOW team for NLPCC2022 shared task in Task5 Track1. The track is about multi-label towards abstracts of academic papers in scientific domain, which includes hierarchical dependencies among 1,530 labels. In order to distinguish semantic information among hierarchical label structures, we propose the Label-aware Graph Convolutional Network (LaGCN), which uses Graph Convolutional Network to capture the label association through context-based label embedding. Besides, curriculum learning is applied for domain adaptation and to mitigate the impact of a large number of categories. The experiments show that: 1) LaGCN effectively models the category information and makes a considerable improvement in dealing with a large number of categories; 2) Curriculum learning is beneficial for a single model in the complex task. Our best results were obtained by an ensemble model. According to the official results, our approach proved the best in this track.",

keywords = "Curriculum learning, Graph convolutional network, Hierarchical multi-label classification, Label embedding",

author = "Bo Wang and Lu, {Yi Fan} and Xiaochi Wei and Xiao Liu and Ge Shi and Changsen Yuan and Heyan huang and Chong Feng and Xianling Mao",

note = "Publisher Copyright: {\textcopyright} 2022, Springer Nature Switzerland AG.; 11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022 ; Conference date: 24-09-2022 Through 25-09-2022",

year = "2022",

doi = "10.1007/978-3-031-17189-5_16",

language = "English",

isbn = "9783031171888",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "192--203",

editor = "Wei Lu and Shujian Huang and Yu Hong and Xiabing Zhou",

booktitle = "Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings",

address = "Germany",

}

Wang, B, Lu, YF, Wei, X, Liu, X, Shi, G, Yuan, C, huang, H , Feng, C & Mao, X 2022, BIT-WOW at NLPCC-2022 Task5 Track1: Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network. in W Lu, S Huang, Y Hong & X Zhou (eds), Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13552 LNAI, Springer Science and Business Media Deutschland GmbH, pp. 192-203, 11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022, Guilin, China, 24/09/22. https://doi.org/10.1007/978-3-031-17189-5_16

BIT-WOW at NLPCC-2022 Task5 Track1: Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network. / Wang, Bo; Lu, Yi Fan; Wei, Xiaochi et al.
Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. ed. / Wei Lu; Shujian Huang; Yu Hong; Xiabing Zhou. Springer Science and Business Media Deutschland GmbH, 2022. p. 192-203 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13552 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - BIT-WOW at NLPCC-2022 Task5 Track1

T2 - 11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022

AU - Wang, Bo

AU - Lu, Yi Fan

AU - Wei, Xiaochi

AU - Liu, Xiao

AU - Shi, Ge

AU - Yuan, Changsen

AU - huang, Heyan

AU - Feng, Chong

AU - Mao, Xianling

PY - 2022

Y1 - 2022

N2 - This paper describes the system proposed by the BIT-WOW team for NLPCC2022 shared task in Task5 Track1. The track is about multi-label towards abstracts of academic papers in scientific domain, which includes hierarchical dependencies among 1,530 labels. In order to distinguish semantic information among hierarchical label structures, we propose the Label-aware Graph Convolutional Network (LaGCN), which uses Graph Convolutional Network to capture the label association through context-based label embedding. Besides, curriculum learning is applied for domain adaptation and to mitigate the impact of a large number of categories. The experiments show that: 1) LaGCN effectively models the category information and makes a considerable improvement in dealing with a large number of categories; 2) Curriculum learning is beneficial for a single model in the complex task. Our best results were obtained by an ensemble model. According to the official results, our approach proved the best in this track.

AB - This paper describes the system proposed by the BIT-WOW team for NLPCC2022 shared task in Task5 Track1. The track is about multi-label towards abstracts of academic papers in scientific domain, which includes hierarchical dependencies among 1,530 labels. In order to distinguish semantic information among hierarchical label structures, we propose the Label-aware Graph Convolutional Network (LaGCN), which uses Graph Convolutional Network to capture the label association through context-based label embedding. Besides, curriculum learning is applied for domain adaptation and to mitigate the impact of a large number of categories. The experiments show that: 1) LaGCN effectively models the category information and makes a considerable improvement in dealing with a large number of categories; 2) Curriculum learning is beneficial for a single model in the complex task. Our best results were obtained by an ensemble model. According to the official results, our approach proved the best in this track.

KW - Curriculum learning

KW - Graph convolutional network

KW - Hierarchical multi-label classification

KW - Label embedding

UR - http://www.scopus.com/inward/record.url?scp=85140775642&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-17189-5_16

DO - 10.1007/978-3-031-17189-5_16

M3 - Conference contribution

AN - SCOPUS:85140775642

SN - 9783031171888

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 192

EP - 203

BT - Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings

A2 - Lu, Wei

A2 - Huang, Shujian

A2 - Hong, Yu

A2 - Zhou, Xiabing

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 24 September 2022 through 25 September 2022

ER -

Wang B, Lu YF, Wei X, Liu X, Shi G, Yuan C et al. BIT-WOW at NLPCC-2022 Task5 Track1: Hierarchical Multi-label Classification via Label-Aware Graph Convolutional Network. In Lu W, Huang S, Hong Y, Zhou X, editors, Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 192-203. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-17189-5_16