Exploration and Optimization of a Deep Reinforcement Learning-based Model for the Creation of Children’s Literature

Authors

  • Jianing Yue College of Humanities and Social Sciences, Heilongjiang Bayi Agricultural University, Daqing 163319, Heilongjiang, China
  • Bing Liu College of Marxism, Heilongjiang Bayi Agricultural University, Daqing 163319, Heilongjiang, China

Abstract

The creation of literature for children is an important research direction in natural language processing and one of the means of improving children’s academic outcomes. This paper explores and optimizes a deep reinforcement learning-based model for the creation of children’s literature; i.e., the technique of deep reinforcement learning is used to generate literary works for children that conform to the characteristics and principles of children’s literature, such as fairy tales, fables, and other fiction. In this paper, a deep reinforcement learning framework based on a generative adversarial network (GAN) is adopted to design a children’s literature creation model consisting of a generator and a discriminator, which is responsible for generating children’s literature. The discriminator is responsible for evaluating the quality of the generated works and giving reward signals to guide the generator to optimize its strategy. The model comprehensively considers the characteristics of theme, style, structure, language and other aspects of children’s literature, and designs a multi-dimensional evaluation index system, including theme relevance, style consistency, structural completeness, language fluency, etc., as well as a comprehensive evaluation index, which is used to measure the overall quality of the generated children’s literature. Moreover, by means of four experiments, this paper tested the ability of the model to generate children’s literature with different themes, styles, structures and lengths, and compared it with random generation, RNN generation, GPT-2 generation, manual generation and other methods. This study aims to provide innovative methods for children’s literature creation by exploring and optimizing models based on deep reinforcement learning, in order to generate more creative, educational and child-friendly story content.

Keywords: deep reinforcement learning, children’s literature creation, text generation

Cite As

J. Yue, B. Liu, "Exploration and Optimization of a Deep Reinforcement Learning-based
Model for the Creation of Children’s Literature", Engineering Intelligent Systems, vol. 32
no. 6, pp. 605-612, 2024.


Downloads

Published

2024-11-01

Issue

Vol. 32 No. 6 (2024): International Journal of Engineering Intelligent Systems

Section

Article

License

The submission of a paper implies that, if accepted for publication, it will not be published elsewhere in the same form, in any language, without the prior consent of the publisher. Before publication, authors are requested to assign copyright to CRL Publishing Ltd. This allows CRL to sanction photocopying, and to authorize the reprinting of issues or volumes according to demand. Authors' traditional rights will not be jeopardized by assigning Copyright in this way, as they retain the right to reuse the material following publication, and to veto third-party publication.