Title: Study on recommendation of personalised learning resources based on deep reinforcement learning

Authors: Zilong Li; Hongdong Wang

Addresses: School of Information Engineering, Xu'zhou University of Technology, Xu'zhou 221018, China ' School of Information Engineering, Xu'zhou University of Technology, Xu'zhou 221018, China

Abstract: In order to overcome the problems of the traditional network personalised learning resource recommendation methods, such as low recommendation accuracy, poor recommendation quality and poor F1 comprehensive evaluation index, a network personalised learning resource recommendation method based on deep reinforcement learning was proposed. This method uses web crawler technology to capture learning resource data. Based on this, a deep reinforcement learning strategy model is built, and the recommended trajectory of network personalised learning resources is divided into independent states. The correlation degree between different network personalised learning resource variables is measured, and the objective function of personalised learning resource recommendation is constructed by combining resource keywords and track segmentation results to complete the resource recommendation. The experiment proves that the accuracy rate of the personalised resources recommended in this paper is above 70%, and the comprehensive evaluation index obviously reaches the highest 99%, which improves the resource recommendation effect.

Keywords: deep reinforcement learning; network; personalised learning resources; recommended.

DOI: 10.1504/IJICT.2023.134832

International Journal of Information and Communication Technology, 2023 Vol.23 No.4, pp.299 - 313

Received: 30 Jun 2021
Accepted: 31 Aug 2021
Published online: 14 Nov 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article