Abstract
Existing work on controlled text generation (CTG) assumes a control interface of categorical attributes. In this work, we propose a natural language (NL) interface, where we craft a PCFG to embed the control attributes into natural language commands, and propose variants of existing CTG models that take commands as input. In our experiments, we design tailored setups to test the model’s generalization abilities. We find our PCFG-based command generation approach is effective for handling unseen commands compared to fix-set templates. Further, our proposed NL models can effectively generalize to unseen attributes (a new ability enabled by the NL interface), as well as unseen attribute combinations. Interestingly, in model comparisons, the simple conditional generation approach, enhanced with our proposed NL interface, is shown to be a strong baseline in those challenging settings.- Anthology ID:
- 2023.starsem-1.27
- Volume:
- Proceedings of the 12th Joint Conference on Lexical and Computational Semantics (*SEM 2023)
- Month:
- July
- Year:
- 2023
- Address:
- Toronto, Canada
- Editors:
- Alexis Palmer, Jose Camacho-collados
- Venue:
- *SEM
- SIG:
- SIGLEX
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 295–313
- Language:
- URL:
- https://aclanthology.org/2023.starsem-1.27
- DOI:
- 10.18653/v1/2023.starsem-1.27
- Cite (ACL):
- Jingyu Zhang, James Glass, and Tianxing He. 2023. PCFG-Based Natural Language Interface Improves Generalization for Controlled Text Generation. In Proceedings of the 12th Joint Conference on Lexical and Computational Semantics (*SEM 2023), pages 295–313, Toronto, Canada. Association for Computational Linguistics.
- Cite (Informal):
- PCFG-Based Natural Language Interface Improves Generalization for Controlled Text Generation (Zhang et al., *SEM 2023)
- PDF:
- https://preview.aclanthology.org/proper-vol2-ingestion/2023.starsem-1.27.pdf