CNN models for readability of Chinese texts

Han Feng; Sizai Hou; Le-Yin Wei; Ding-Xuan Zhou

doi:10.3934/mfc.2022021

Article Contents

2022, Volume 5, Issue 4: 351-362. Doi: 10.3934/mfc.2022021

This issue Previous Article A generalized projection iterative method for solving non-singular linear systems Next Article

CNN models for readability of Chinese texts

1.
Department of Mathematics, City University of Hong Kong, Hong Kong
2.
The Hong Kong University of Science and Technology (Guangzhou), Nansha, Guangzhou, 511400, Guangdong, China
3.
Shaoxing University, Shaoxing 312000, Zhejiang, China
4.
School of Data Science, Department of Mathematics, Liu Bie Ju Centre for Mathematical Sciences, City University of Hong Kong, Hong Kong

^*Corresponding author: Le-Yin Wei
^*Corresponding author: Le-Yin Wei

Received: June 2022

Early access: July 2022

Published: November 2022

Abstract / Introduction Full Text(HTML) Figure(5) / Table(2) Related Papers Cited by

Abstract

Readability of Chinese texts considered in this paper is a multi-class classification problem with $ 12 $ grade classes corresponding to $ 6 $ grades in primary schools, $ 3 $ grades in middle schools, and $ 3 $ grades in high schools. A special property of this problem is the strong ambiguity in determining the grades. To overcome the difficulty, a measurement of readability assessment methods used empirically in practice is adjacent accuracy in addition to exact accuracy. In this paper we give mathematical definitions of these concepts in a learning theory framework and compare these two quantities in terms of the ambiguity level of texts. A deep learning algorithm is proposed for readability of Chinese texts, based on convolutional neural networks and a pre-trained BERT model for vector representations of Chinese characters. The proposed CNN model can extract sentence and text features by convolutions of sentence representations with filters and is efficient for readability assessment, which is demonstrated with some numerical experiments.

Keywords:

Mathematics Subject Classification: Primary: 68T07, 68T50; Secondary: 68Q32.

Citation:

Full Text(HTML)

Figure 1. One Filter Instance

Download: Full-size image PowerPoint slide

Figure 2. Two Inputs & Two Filters

Download: Full-size image PowerPoint slide

Figure 3. Accuracy Curve by Epoch Number

Download: Full-size image PowerPoint slide

Figure 4. Confusion Matrix

Download: Full-size image PowerPoint slide

Figure 5. Scatter Plot

Download: Full-size image PowerPoint slide

Table 1. Number of Texts in Each Grade

Grade	1	2	3	4	5	6	7	8	9	10	11	12	Total
Texts	235	320	386	321	281	252	145	58	134	86	26	109	2353

| Show Table

DownLoad: CSV

Table 2. Empirical accuracies of various models

Model	Vec2Read[12]	Tseng et al.[12]	Basic	Multi-Channel	Top-$ k $	Fused
$ \hat{{\mathcal A}} $	29.18	29.00	43.9	44.8	45.6	48.6
$ \hat{{\mathcal A}}_{\mathcal C} $	69.70	67.05	83.7	84.3	83.7	88.0

| Show Table

DownLoad: CSV

Related Papers

Cited by

References

[1]	D. Chen and D. C. Manning, A fast and accurate dependency parser using neural networks, Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), (2014), 740–750. doi: 10.3115/v1/D14-1082.
[2]	T. Cohn, Y. He and Y. Liu, Glove: Global vectors for word representation, Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), (2014), 1532–1543.
[3]	E. Dale and J. S. Chall, The concept of readability, Elementary English, 26 (1949), 19-26.
[4]	J. Devlin, M. W. Chang, K. Lee and K. Toutanova, Bert: Pre-training of deep bidirectional transformers for language understanding, Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, (2019), 4171–4186.
[5]	Z. Fang, H. Feng, S. Huang and D. X. Zhou, Theory of deep convolutional neural networks II: Spherical analysis, Neural Networks, 131 (2020), 154-162.
[6]	H. Feng, S. Huang and D. X. Zhou, Generalization analysis of CNNs for classification on spheres, IEEE Transactions on Neural Networks and Learning Systems, in press.
[7]	J. R. Firth, A synopsis of linguistic theory 1930-55, Studies in Linguistic Analysis (Special Volume of the Philological Society), The Philological Society, (1957), 1–32.
[8]	Y. Kim, Convolutional neural networks for sentence classification, Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), (2014), 1746–1751.
[9]	A. Krizhevsky, I. Sutskever and E. Hinton, Imagenet classification with deep convolutional neural networks, Advances in Neural Information Processing Systems, 25 (2012), 1097-1105.
[10]	Y. LeCun, L. Bottou, Y. Bengio and P. Haffner, Gradient-based learning applied to document recognition, Proceedings of the IEEE, 86 (1998), 2278-2324. doi: 10.1109/5.726791.
[11]	T. Mikolov, K. Chen, G. Corrado and J. Dean, Efficient estimation of word representations in vector space, ICLR, 2013.
[12]	J. Zeng, Y. Xie, J. Lee and D. X. Zhou, CR-BERT: Chinese text readability method based on BERT and multi-level attention, preprint, (2022).
[13]	D. X. Zhou, Universality of deep convolutional neural networks, Appl. Comput. Harmon. Anal., 48 (2020), 787-794. doi: 10.1016/j.acha.2019.06.004.
[14]	D. X. Zhou, Deep distributed convolutional neural networks: Universality, Analysis and Applications, 16 (2018), 895-919. doi: 10.1142/S0219530518500124.
[15]	D. X. Zhou, Theory of deep convolutional neural networks: Downsampling, Neural Networks, 124 (2020), 319-327.