%A Cheng Quan, Dong Jia %T Hierarchical Multi-label Classification of Children's books for Graded Reading %0 Journal Article %D 0 %J Data Analysis and Knowledge Discovery %R 10.11925/infotech.2096-3467.2022-0649 %P 1- %V %N %U {https://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/abstract/article_5501.shtml} %8 %X

[Objective] In order to realize the automatic classification of children's books,a hierarchical multi-label classification model of children's books is constructed to guide children readers to choose books that are more suitable for their own development. [Methods]The concept of graded reading is embodied into a hierarchical classification system, a hierarchical multi-label text classification model based on ERNIE-HAM is constructed by deep learning technology. [Results]By comparing the four pre-trained models, the ERNIE-HAM model has better performance in the second and third layers of the hierarchical classification of children's books; comparing the single layer algorithm, the hierarchical algorithm improves the  by about 11% in both the second and third layers; comparing the two hierarchical multi-label classification models, HFT-CNN and HMCN, the ERNIE-HAM model has the third layer improved by 12.79% and 6.48% in the classification results, respectively. [Limitations]The overall classification results of the model have not yet reached expectations, and more refinement and exploration can be done in the future in terms of the expansion of the data set and algorithm design. [Conclusions]The effectiveness of the ERNIE-HAM model proposed in this study on the hierarchical multi-label classification task of children's reading materials was verified through three sets of comparison experiments.