Improved Unsupervised Chinese Word Segmentation Using Pre-trained Knowledge and Pseudo-labeling Transfer
Hsiu-Wen Li, Ying-Jia Lin, Yi-Ting Li, Chun Lin, Hung-Yu Kao
Abstract
Unsupervised Chinese word segmentation (UCWS) has made progress by incorporating linguistic knowledge from pre-trained language models using parameter-free probing techniques. However, such approaches suffer from increased training time due to the need for multiple inferences using a pre-trained language model to perform word segmentation. This work introduces a novel way to enhance UCWS performance while maintaining training efficiency. Our proposed method integrates the segmentation signal from the unsupervised segmental language model to the pre-trained BERT classifier under a pseudo-labeling framework. Experimental results demonstrate that our approach achieves state-of-the-art performance on the eight UCWS tasks while considerably reducing the training time compared to previous approaches.- Anthology ID:
- 2023.emnlp-main.564
- Volume:
- Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
- Month:
- December
- Year:
- 2023
- Address:
- Singapore
- Editors:
- Houda Bouamor, Juan Pino, Kalika Bali
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 9109–9118
- Language:
- URL:
- https://aclanthology.org/2023.emnlp-main.564
- DOI:
- 10.18653/v1/2023.emnlp-main.564
- Cite (ACL):
- Hsiu-Wen Li, Ying-Jia Lin, Yi-Ting Li, Chun Lin, and Hung-Yu Kao. 2023. Improved Unsupervised Chinese Word Segmentation Using Pre-trained Knowledge and Pseudo-labeling Transfer. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 9109–9118, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- Improved Unsupervised Chinese Word Segmentation Using Pre-trained Knowledge and Pseudo-labeling Transfer (Li et al., EMNLP 2023)
- PDF:
- https://preview.aclanthology.org/emnlp-22-attachments/2023.emnlp-main.564.pdf