JACIII Vol.23 No.4 pp. 719-725
doi: 10.20965/jaciii.2019.p0719


Ancient Chinese Sentence Segmentation Based on Bidirectional LSTM+CRF Model

Hongbin Wang*, Haibing Wei*, Jianyi Guo*, and Liang Cheng**,†

*Faculty of Information Engineering and Automation, Kunming University of Science and Technology
No.727 Jingming South Road, Chenggong New Area, Kunming, Yunnan 650504, China

**City College, Kunming University of Science and Technology
No.50 East Ring Road, Kunming, Yunnan 650051, China

Corresponding author

May 29, 2018
February 18, 2019
July 20, 2019
long short-term memory, CRF, ancient chinese, sentence segmentation

This study proposes a novel method for the segmentation of Archaic Chinese sentences based on a bidirectional long short-term memory (LSTM) + conditional random field (CRF) model. The method added a layer of linear statistical model to the traditional bidirectional LSTM neural network; it can be used for sequence annotation from the sentence level. In addition, this model introduced the stochastic gradient descent (SGD) to prevent excessive fitting, and the viterbi algorithm was used to calculate the optimal sequence of the sentences. In the experiment, this study tests the performance of the proposed method using the History of the Han Dynasty, the History of the later Han Dynasty, Three Kingdoms, and the Book of Jin, amongst others. The results show that the precision value, recall value, and F1 value are 0.77, 0.75, and 0.76, respectively, in the open test, and 0.90, 0.88, and 0.76, respectively, in the closed test.

Cite this article as:
H. Wang, H. Wei, J. Guo, and L. Cheng, “Ancient Chinese Sentence Segmentation Based on Bidirectional LSTM+CRF Model,” J. Adv. Comput. Intell. Intell. Inform., Vol.23, No.4, pp. 719-725, 2019.
Data files:
  1. [1] N. Ketui, T. Theeramunkong, and C. Onsuwan, “A Rule-Based Method for Thai Elementary Discourse Unit Segmentation (TED-Seg),” Proc. of 7th Int. Conf. on Knowledge, Information and Creativity Support Systems, pp. 195-202, 2012.
  2. [2] L. Liu, H. Shi, and Y. Zhang, “A new method of phrase segmentation in statistical machine translation,” Electronic Test, No.2, pp. 26-27, 2017 (in Chinese).
  3. [3] B. Wang, X. Shi, and J. Su, “A Sentence Segmentation Method for Ancient Chinese Texts Based on Recurrent Neural Network,” Acta Scientiarum Naturalium Universitatis Pekinensis, Vol.53, No.2, pp. 255-261, 2017 (in Chinese).
  4. [4] S.-L. Yang, “The Study on Sentence Segmentation for English-Chinese Machine Translation,” Master Thesis, Beijing Institute of Technology, 2016 (in Chinese).
  5. [5] T. Chen, R. Chen, L. Pan, H. Li, and Z. Yu, “Archaic Chinese Punctuating Sentences Based on Context N-gram Model,” Computer Engineering, Vol.33, No.3, pp. 192-193, 2007 (in Chinese).
  6. [6] C. Wang, X. Zhang, and C. Han, “Research on Sentence Segmentation and Punctuation in Ancient Chinese,” J. of Henan University (Natural Science), Vol.39, No.5, pp. 525-529, 2009 (in Chinese).
  7. [7] Y. Kong, Z. Wang, and L. Che, “Real-time Face Recognition in Videos Based on Convolutional Neural Networks (CNN) and CUDA,” Science Technology and Engineering, Vol.16, No.35, pp. 96-100+107, 2016 (in Chinese).
  8. [8] L. Huang and C. Du, “Application of recurrent neural networks in text classification,” J. of Beijing University of Chemical Technology (Natural Science), Vol.44, Issue 1, pp. 98-104, 2017 (in Chinese).
  9. [9] X. Hu, “Research on Semantic Relation Classification Based on LSTM,” Master thesis, Harbin Institute of Technology, 2015 (in Chinese).
  10. [10] X. Li, H. Duan, and M. Xu, “A Gated Recurrent Unit Neural Network for Chinese Word Segmentation,” J. of Xiamen University (Natural Science), Vol.56, No.2, pp. 237-243, 2017 (in Chinese).
  11. [11] Z. Ren, H. Xu, S. Feng, H. Zhou, and J. Shi, “Sequence labeling Chinese word segmentation method based on LSTM networks,” Application Research of Computers, Vol.34, No.5, pp. 1321-1324+1341, 2017 (in Chinese).
  12. [12] J. Huang, “Chinese Word Segmentation Analysis based on Bidirectional LSTMN Recurrent Neural Network,” Master thesis, Nanjing University, 2016 (in Chinese).
  13. [13] Z. Huang, W. Xu, and K. Yu, “Bidirectional LSTM-CRF Models for Sequence Tagging,” arXiv: 1508.01991, 2015.
  14. [14] G.-B. Zhou, J. Wu, C.-L. Zhang, and Z.-H. Zhou, “Minimal gated unit for recurrent neural networks,” Int. J. of Automation and Computing, Vol.13, Issue 3, pp. 226-234, 2016.
  15. [15] G. Lample, M. Ballesteros, S. Subramanian, K. Kawakami, and C. Dyer, “Neural Architectures for Named Entity Recognition,” Proc. of NAACL-HLT, pp. 260-270, 2016.
  16. [16] S. Ghosh, O. Vinyals, B. Strope, et al., “Contextual LSTM (CLSTM) models for Large scale NLP tasks,” arXiv: 1602.06291, 2016.
  17. [17] [accessed April 20, 2018]

*This site is desgined based on HTML5 and CSS3 for modern browsers, e.g. Chrome, Firefox, Safari, Edge, IE9,10,11, Opera.

Last updated on Sep. 19, 2019