A Novel Perspective to Look At Attention: Bi-level Attention-based Explainable Topic Modeling for News Classification

Dairui Liu, Derek Greene, Ruihai Dong


Abstract
Many recent deep learning-based solutions have adopted the attention mechanism in various tasks in the field of NLP. However, the inherent characteristics of deep learning models and the flexibility of the attention mechanism increase the models’ complexity, thus leading to challenges in model explainability. To address this challenge, we propose a novel practical framework by utilizing a two-tier attention architecture to decouple the complexity of explanation and the decision-making process. We apply it in the context of a news article classification task. The experiments on two large-scaled news corpora demonstrate that the proposed model can achieve competitive performance with many state-of-the-art alternatives and illustrate its appropriateness from an explainability perspective. We release the source code here.
Anthology ID:
2022.findings-acl.178
Volume:
Findings of the Association for Computational Linguistics: ACL 2022
Month:
May
Year:
2022
Address:
Dublin, Ireland
Editors:
Smaranda Muresan, Preslav Nakov, Aline Villavicencio
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2280–2290
Language:
URL:
https://aclanthology.org/2022.findings-acl.178
DOI:
10.18653/v1/2022.findings-acl.178
Bibkey:
Cite (ACL):
Dairui Liu, Derek Greene, and Ruihai Dong. 2022. A Novel Perspective to Look At Attention: Bi-level Attention-based Explainable Topic Modeling for News Classification. In Findings of the Association for Computational Linguistics: ACL 2022, pages 2280–2290, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
A Novel Perspective to Look At Attention: Bi-level Attention-based Explainable Topic Modeling for News Classification (Liu et al., Findings 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/dois-2013-emnlp/2022.findings-acl.178.pdf
Video:
 https://preview.aclanthology.org/dois-2013-emnlp/2022.findings-acl.178.mp4
Code
 ruixinhua/batm +  additional community code
Data
MIND