Large-scale text pre-training helps with dialogue act recognition, but not without fine-tuning

Bill Noble, Vladislav Maraev


Abstract
We use dialogue act recognition (DAR) to investigate how well BERT represents utterances in dialogue, and how fine-tuning and large-scale pre-training contribute to its performance. We find that while both the standard BERT pre-training and pretraining on dialogue-like data are useful, task-specific fine-tuning is essential for good performance.
Anthology ID:
2021.iwcs-1.16
Volume:
Proceedings of the 14th International Conference on Computational Semantics (IWCS)
Month:
June
Year:
2021
Address:
Groningen, The Netherlands (online)
Venue:
IWCS
SIG:
SIGSEM
Publisher:
Association for Computational Linguistics
Note:
Pages:
166–172
Language:
URL:
https://aclanthology.org/2021.iwcs-1.16
DOI:
Bibkey:
Cite (ACL):
Bill Noble and Vladislav Maraev. 2021. Large-scale text pre-training helps with dialogue act recognition, but not without fine-tuning. In Proceedings of the 14th International Conference on Computational Semantics (IWCS), pages 166–172, Groningen, The Netherlands (online). Association for Computational Linguistics.
Cite (Informal):
Large-scale text pre-training helps with dialogue act recognition, but not without fine-tuning (Noble & Maraev, IWCS 2021)
Copy Citation:
PDF:
https://preview.aclanthology.org/paclic-22-ingestion/2021.iwcs-1.16.pdf
Data
OpenSubtitles