Team JUSTR00 at SemEval-2023 Task 3: Transformers for News Articles Classification

Ahmed Al-Qarqaz, Malak Abdullah


Abstract
The SemEval-2023 Task 3 competition offers participants a multi-lingual dataset with three schemes one for each subtask. The competition challenges participants to construct machine learning systems that can categorize news articles based on their nature and style of writing. We esperiment with many state-of-the-art transformer-based language models proposed in the natural language processing literature and report the results of the best ones. Our top performing model is based on a transformer called “Longformer” and has achieved an F1-Micro score of 0.256 on the English version of subtask-1 and F1-Macro of 0.442 on subtask-2 on the test data. We also experiment with a number of state-of-the-art multi-lingual transformer-based models and report the results of the best performing ones.
Anthology ID:
2023.semeval-1.168
Volume:
Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023)
Month:
July
Year:
2023
Address:
Toronto, Canada
Editors:
Atul Kr. Ojha, A. Seza Doğruöz, Giovanni Da San Martino, Harish Tayyar Madabushi, Ritesh Kumar, Elisa Sartori
Venue:
SemEval
SIG:
SIGLEX
Publisher:
Association for Computational Linguistics
Note:
Pages:
1213–1216
Language:
URL:
https://aclanthology.org/2023.semeval-1.168
DOI:
10.18653/v1/2023.semeval-1.168
Bibkey:
Cite (ACL):
Ahmed Al-Qarqaz and Malak Abdullah. 2023. Team JUSTR00 at SemEval-2023 Task 3: Transformers for News Articles Classification. In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 1213–1216, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
Team JUSTR00 at SemEval-2023 Task 3: Transformers for News Articles Classification (Al-Qarqaz & Abdullah, SemEval 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.semeval-1.168.pdf