Abstract
In 2019, Reimers et al. proposed SBERT to derive sentence embedding for many purposes. It highly reduced the time complexity of finding the most similar pair of sentences from traditional BERT/RoBERTa to SBERT, while the accuracy is maintained. There are many English SBERT models, but lacking the other languages ones. In this publication, we develop our Vietnamese SBERT model for Vietnamese sentence embeddings, using PhoBERT as our main transformer for Vietnamese token embeddings. For the training processes, we use the Vietnamese NLI and STSb datasets, and for the evaluation of sentence paraphrase identification task to compare with other models, we use the VnPara dataset in. Our model has achieved an accuracy of 95.33% and F1 of 95.42%, slightly outperforming many recent methods in Vietnamese.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
Our Vietnamese SBERT model: https://huggingface.co/keepitreal/vietnamese-sbert.
References
Reimers, N., Gurevych, I.: Sentence-BERT: sentence embeddings using siamese BERT-networks. In: EMNLP 2019 (2019). arXiv:1908.10084
Nguyen, D.Q., Nguyen, A.T.: PhoBERT: pre-trained language models for Vietnamese, pp. 1037–1042. Association for Computational Linguistics (2020). https://doi.org/10.18653/v1/2020.findings-emnlp.92
Dinh, D., Le Thanh, N.: Vietnamese sentence paraphrase identification using pre-trained model and linguistic knowledge. Int. J. Adv. Comput. Sci. Appl. (IJACSA) 12(8), 2021 (2021). https://doi.org/10.14569/IJACSA.2021.0120891
Vani, K., Gupta, D.: Study on extrinsic text plagiarism detection techniques and tools. J. Eng. Sci. Technol. Rev. (2016). https://doi.org/10.25103/jestr.095.02
Devlin, J., Chang, M.-W., Lee, K., Toutanova, K.: BERT: pre-training of deep bidirectional transformers for language understanding. arXiv:1810.04805v2 [cs.CL], 24 May 2019
Vaswani, A., et al.: Attention is all you need. arXiv:1706.03762v5 [cs.CL], 6 December 2017
Conneau, A., et al.: Unsupervised cross-lingual representation learning at scale. arXiv:1911.02116 [cs.CL], 5 November 2019
Liu, Y., et al: RoBERTa: a robustly optimized BERT pretraining approach. arXiv:2003.00744 [cs.CL], 26 July 2019
Bach, N.X., Oanh, T., Hai, N., Phuong, T.: Paraphrase identification in Vietnamese documents. In: 2015 IEEE International Conference on Knowledge and Systems Engineering, KSE 2015, pp. 174–179 (2015). ISBN: 9781467380133. https://doi.org/10.1109/KSE.2015.37
SQuAD2.0 The Stanford Question Answering Dataset. https://rajpurkar.github.io/SQuAD-explorer/
Vietnamese NLI Dataset, Dat Quoc Nguyen. https://github.com/DatCanCode/sentence-transformers/tree/master/DataNLI
Semantic Textual Similarity Wiki. http://ixa2.si.ehu.eus/stswiki
Acknowledgement
We are very grateful to our instructors who helped us review our works and our friends for their valuable support. Also, we acknowledge the support of facilities from Ho Chi Minh City University of Technology (HCMUT), VNU-HCM for this study.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Phan, Q.L., Doan, T.H.P., Le, N.H., Tran, N.B.D., Huynh, T.N. (2022). Vietnamese Sentence Paraphrase Identification Using Sentence-BERT and PhoBERT. In: Nguyen, NT., Dao, NN., Pham, QD., Le, H.A. (eds) Intelligence of Things: Technologies and Applications . ICIT 2022. Lecture Notes on Data Engineering and Communications Technologies, vol 148. Springer, Cham. https://doi.org/10.1007/978-3-031-15063-0_40
Download citation
DOI: https://doi.org/10.1007/978-3-031-15063-0_40
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-15062-3
Online ISBN: 978-3-031-15063-0
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)