Title: Recommendation system based on improved graph neural networks

Authors: Jiawen Chen; Chao Cai; Yong Cai; Fangbin Yan; Jiayi Li

Addresses: State Grid Hubei Electric Power Co., Ltd., Beijing, China ' State Grid Hubei Electric Power Co., Ltd., Beijing, China ' State Grid Hubei Electric Power Co., Ltd., Beijing, China ' State Grid Hubei Electric Power Co., Ltd., Beijing, China ' School of Electrical and Automation, Wuhan University, Wuhan, Hubei, China

Abstract: This study introduces WideGCN, a novel recommendation system that goes beyond historical user-product interactions to incorporate user and product side information. Unlike traditional NGCF and LGCN methods, WideGCN uses two fully connected neural networks to extract features for users and products, respectively. These features are then integrated with representations learned from LightGCN to produce final user and product representations. Compared to baselines like Popularity, BPR, NGCF, and LGCN, WideGCN consistently achieves higher validation scores and faster training speeds, demonstrating the effectiveness of incorporating side information for enhanced recommendation performance.

Keywords: graph neural networks; wide graph neural networks; recommendation system; user and products features.

DOI: 10.1504/IJWMC.2024.141450

International Journal of Wireless and Mobile Computing, 2024 Vol.27 No.3, pp.290 - 296

Received: 24 Nov 2023
Accepted: 17 Jan 2024
Published online: 13 Sep 2024 *

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