Title: An optimised graph recurrent neural network with chimp explored equilibrium optimisation-based e-commerce recommendation system

Authors: M. Balamurugan; A. Bharathiraja

Addresses: School of Computer Science, Engineering & Applications, Bharathidasan University, Tiruchirappalli, 620-023, India ' School of Computer Science, Engineering & Applications, Bharathidasan University, Tiruchirappalli, 620-023, India

Abstract: Content-based recommendation systems (RS) in e-commerce leverage the characteristics and attributes of products to make personalised recommendations to users. While computing the similarity between user profiles and product attributes in a content-based RS, several problems can arise. To overcome this, the paper presents an optimised graph recurrent neural network (GRNN) with chimp-explored equilibrium optimisation (CEEO). Data for the RS is collected from the Myntra fashion product dataset, pre-processed and a pre-trained AlexNet model is used for image feature extraction, and feature selection using statistical measures and CEEO algorithm. Fine-tuning of the GRNN is done using the Chimp optimisation algorithm (COA) with considerations for parameters such as mini-batch, learning rate, dropout, and momentum. The equilibrium optimisation algorithm fine-tunes the GRNN's weight and bias. The CEEO algorithm fine-tunes all parameters of the GRNN. Experimental results demonstrate the proposed model's superiority with a low error number indicating its efficiency in achieving successful prediction.

Keywords: e-commerce RS; chimp explored equilibrium optimisation; CEEO; OGRNN; review prediction; query-based image recommendation.

DOI: 10.1504/IJMOR.2025.150929

International Journal of Mathematics in Operational Research, 2025 Vol.32 No.4, pp.511 - 535

Received: 30 Sep 2023
Accepted: 07 Oct 2023
Published online: 05 Jan 2026 *

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