Enhancing the Transformer Decoder with Transition-based Syntax

Leshem Choshen, Omri Abend


Abstract
Notwithstanding recent advances, syntactic generalization remains a challenge for text decoders. While some studies showed gains from incorporating source-side symbolic syntactic and semantic structure into text generation Transformers, very little work addressed the decoding of such structure. We propose a general approach for tree decoding using a transition-based approach. Examining the challenging test case of incorporating Universal Dependencies syntax into machine translation, we present substantial improvements on test sets that focus on syntactic generalization, while presenting improved or comparable performance on standard MT benchmarks. Further qualitative analysis addresses cases where syntactic generalization in the vanilla Transformer decoder is inadequate and demonstrates the advantages afforded by integrating syntactic information.
Anthology ID:
2022.conll-1.27
Volume:
Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL)
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates (Hybrid)
Editors:
Antske Fokkens, Vivek Srikumar
Venue:
CoNLL
SIG:
SIGNLL
Publisher:
Association for Computational Linguistics
Note:
Pages:
384–404
Language:
URL:
https://aclanthology.org/2022.conll-1.27
DOI:
10.18653/v1/2022.conll-1.27
Bibkey:
Cite (ACL):
Leshem Choshen and Omri Abend. 2022. Enhancing the Transformer Decoder with Transition-based Syntax. In Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL), pages 384–404, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics.
Cite (Informal):
Enhancing the Transformer Decoder with Transition-based Syntax (Choshen & Abend, CoNLL 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.conll-1.27.pdf
Video:
 https://aclanthology.org/2022.conll-1.27.mp4