Abstract
We participated in the Shared Task 1 at CASE 2021, Subtask 4 on protest event extraction from news articles and examined different techniques aimed at improving the performance of the winning system from the last competition round. We evaluated in-domain pre-training, task-specific pre-fine-tuning, alternative loss function, translation of the English training dataset into other target languages (i.e., Portuguese, Spanish, and Hindi) for the token classification task, and a simple data augmentation technique by random sentence reordering. This paper summarizes the results, showing that random sentence reordering leads to a consistent improvement of the model performance.- Anthology ID:
- 2022.case-1.27
- Volume:
- Proceedings of the 5th Workshop on Challenges and Applications of Automated Extraction of Socio-political Events from Text (CASE)
- Month:
- December
- Year:
- 2022
- Address:
- Abu Dhabi, United Arab Emirates (Hybrid)
- Editors:
- Ali Hürriyetoğlu, Hristo Tanev, Vanni Zavarella, Erdem Yörük
- Venue:
- CASE
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 189–194
- Language:
- URL:
- https://aclanthology.org/2022.case-1.27
- DOI:
- 10.18653/v1/2022.case-1.27
- Cite (ACL):
- Arthur Müller and Andreas Dafnos. 2022. SPARTA at CASE 2021 Task 1: Evaluating Different Techniques to Improve Event Extraction. In Proceedings of the 5th Workshop on Challenges and Applications of Automated Extraction of Socio-political Events from Text (CASE), pages 189–194, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics.
- Cite (Informal):
- SPARTA at CASE 2021 Task 1: Evaluating Different Techniques to Improve Event Extraction (Müller & Dafnos, CASE 2022)
- PDF:
- https://preview.aclanthology.org/emnlp-22-attachments/2022.case-1.27.pdf