Journal
NATURAL LANGUAGE PROCESSING AND CHINESE COMPUTING, NLPCC 2022, PT II
Volume 13552, Issue -, Pages 192-203Publisher
SPRINGER INTERNATIONAL PUBLISHING AG
DOI: 10.1007/978-3-031-17189-5_16
Keywords
Hierarchical multi-label classification; Graph convolutional network; Curriculum learning; Label embedding
Categories
Funding
- Joint Funds of the National Natural Science Foundation of China [U19B2020]
Ask authors/readers for more resources
This paper presents the system developed by the BIT-WOW team for the NLPCC2022 shared task in Task5 Track1. The system utilizes the Label-aware Graph Convolutional Network (LaGCN) to address the multi-label classification task for academic paper abstracts in the scientific domain. The experiments demonstrate the effectiveness of LaGCN in modeling category information and dealing with a large number of categories. Furthermore, the application of curriculum learning contributes to the adaptability of the system. The ensemble model achieved the best performance according to the official results.
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.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available