Deep Learning-Based Recommender Systems: A Performance Evaluation using CAMEL Approach

Abstract

With the explosive growth of digital data and an increasing number of users on the internet, recommender systems have become indispensable in filtering relevant information from massive datasets. Traditional recommender approaches face challenges of scalability and efficiency in handling big data. Deep learning methods have emerged as effective alternatives due to their capability of automated feature extraction, non-linear modeling and handling large-scale data efficiently. This paper evaluates the performance of deep learning-based recommender systems (RS) by analyzing multiple metrics such as accuracy, execution time, latency, jitter, network bandwidth consumption and power consumption. Using Python (Keras API) and real-world datasets, the study demonstrates that deep learning models, especially multilayer perceptrons with backpropagation and momentum, achieve superior results in terms of convergence speed and prediction accuracy. The findings reinforce the potential of deep learning in advancing recommender systems for domains like healthcare, e-commerce and multimedia services.