Image Based Similar Product Recommendation System For E-commerce Platform

Abstract

Research on online product recommendations is crucial, as text-based search functionality has limitations. Image-based similarity search offers enhanced visual search experiences, benefiting e-commerce platforms. A machine-learning-based image-based recommender system was designed using Singular Value Decomposition (SVD) and Principal Component Analysis (PCA) to reduce dimensionality and enhance computational efficiency. The K-Means++ clustering method identified similar product groups, refining the recommendation process. 
The study compared four unsupervised clustering algorithms on a dataset of 44,446 product images from the Kaggle platform. The PCA-SVD transformed K-means++ methodology showed higher performance, delivering higher recommendations compared to other strategies, indicating its potential for revolutionizing e-commerce product recommendations.