HiPool: Modeling Long Documents Using Graph Neural Networks

Irene Li, Aosong Feng, Dragomir Radev, Rex Ying

Abstract
Encoding long sequences in Natural Language Processing (NLP) is a challenging problem. Though recent pretraining language models achieve satisfying performances in many NLP tasks, they are still restricted by a pre-defined maximum length, making them challenging to be extended to longer sequences. So some recent works utilize hierarchies to model long sequences. However, most of them apply sequential models for upper hierarchies, suffering from long dependency issues. In this paper, we alleviate these issues through a graph-based method. We first chunk the sequence with a fixed length to model the sentence-level information. We then leverage graphs to model intra- and cross-sentence correlations with a new attention mechanism. Additionally, due to limited standard benchmarks for long document classification (LDC), we propose a new challenging benchmark, totaling six datasets with up to 53k samples and 4034 average tokens’ length. Evaluation shows our model surpasses competitive baselines by 2.6% in F1 score, and 4.8% on the longest sequence dataset. Our method is shown to outperform hierarchical sequential models with better performance and scalability, especially for longer sequences.
Anthology ID:
2023.acl-short.16
Volume:
Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)
Month:
July
Year:
2023
Address:
Toronto, Canada
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
161–171
Language:
URL:
https://aclanthology.org/2023.acl-short.16
DOI:
10.18653/v1/2023.acl-short.16
Bibkey:
Cite (ACL):
Irene Li, Aosong Feng, Dragomir Radev, and Rex Ying. 2023. HiPool: Modeling Long Documents Using Graph Neural Networks. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 161–171, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
HiPool: Modeling Long Documents Using Graph Neural Networks (Li et al., ACL 2023)
Copy Citation:
PDF:
https://preview.aclanthology.org/remove-xml-comments/2023.acl-short.16.pdf
PDF Search