HCMUS_The Fangs at AbjadStyleTransfer Shared Task: Learning to Query Style, Contrastive Representations for Zero-Shot Arabic Authorship Style Transfer

Duy Minh Dao Sy, Trung Kiet Huynh, Nguyen Chi Tran, Nguyen Lam Phu Quy, Pham Phu Hoa, Nguyen Dinh Ha Duong


Abstract
This paper describes the system developed by team HCMUS_The Fangs for the AbjadStyleTransfer shared task (ArabicNLP 2026), where we achieved 1st place. We present a contrastive style learning approach for zero-shot Arabic authorship style transfer. Our key discovery is that the 21 test authors-including Nobel laureate Naguib Mahfouz and literary pioneer Taha Hussein-have zero overlap with the 32,784 training authors, transforming this into a pure zero-shot challenge. This insight led us to develop a dual-encoder architecture that learns transferable style representations through contrastive objectives, rather than memorizing author-specific patterns. Our system achieves 19.77 BLEU and 55.74 chrF, outperforming retrieval-augmented generation (+18%) and multi-task learning (+31%). Counter-intuitively, we find that sophisticated architectural modifications like style injection consistently degrade performance, while simpler approaches that preserve pre-trained knowledge excel. Our analysis reveals that for famous authors, pre-trained Arabic language models already encode substantial stylistic knowledge-the key is surfacing it, not learning from scratch.
Anthology ID:
2026.abjadnlp-1.52
Volume:
Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script
Month:
March
Year:
2026
Address:
Rabat, Morocco
Venues:
AbjadNLP | WS
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
438–442
Language:
URL:
https://aclanthology.org/2026.abjadnlp-1.52/
DOI:
Bibkey:
Cite (ACL):
Duy Minh Dao Sy, Trung Kiet Huynh, Nguyen Chi Tran, Nguyen Lam Phu Quy, Pham Phu Hoa, and Nguyen Dinh Ha Duong. 2026. HCMUS_The Fangs at AbjadStyleTransfer Shared Task: Learning to Query Style, Contrastive Representations for Zero-Shot Arabic Authorship Style Transfer. In Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script, pages 438–442, Rabat, Morocco. Association for Computational Linguistics.
Cite (Informal):
HCMUS_The Fangs at AbjadStyleTransfer Shared Task: Learning to Query Style, Contrastive Representations for Zero-Shot Arabic Authorship Style Transfer (Dao Sy et al., AbjadNLP 2026)
Copy Citation:
PDF:
https://aclanthology.org/2026.abjadnlp-1.52.pdf