CNN models for readability of Chinese texts

<p style='text-indent:20px;'>Readability of Chinese texts considered in this paper is a multi-class classification problem with <inline-formula><tex-math id="M1">\begin{document}$ 12 $\end{document}</tex-math></inline-formula> grade classes corresponding to id="M2">\begin{document}$ 6 grades primary schools, id="M3">\begin{document}$ 3 middle and id="M4">\begin{document}$ high schools. A special property the strong ambiguity determining grades. To overcome difficulty, measurement readability assessment methods used empirically practice adjacent accuracy addition exact accuracy. In we give mathematical definitions these concepts learning theory framework compare two quantities terms level texts. deep algorithm proposed for texts, based on convolutional neural networks pre-trained BERT model vector representations characters. The CNN can extract sentence text features by convolutions filters efficient assessment, which demonstrated some numerical experiments.</p>