Hybrid Movie Recommendation Engine Using Weighted Classification and Collaborative Filtering

Abstract

The global video-on-demand market is projected to surpass $257 billion by 2027, with platforms like 
Netflix attributing over 80% of content watched to recommendation engines. This highlights the 
growing demand for intelligent, accurate, and personalized movie recommendation systems. However, 
existing systems face persistent issues such as data sparsity, cold-start problems, and lack of content 
diversity, which hinder their effectiveness and scalability. To address these limitations, this research 
introduces a novel hybrid recommendation framework named CBCF-CNN (Content-Based 
Collaborative Filtering integrated with Convolutional Neural Networks). The proposed model 
innovatively combines the strengths of content-based filtering, collaborative filtering, and deep learning 
to deliver a robust and scalable recommendation system. CBCF-CNN leverages user-item interaction 
matrices along with high-dimensional feature extraction from movie metadata and visual content using 
CNNs. This deep integration allows the system to capture latent patterns and semantic similarities 
between items, thus mitigating the cold-start issue and enriching the recommendation space. 
Furthermore, the architecture enhances personalization by learning user-specific behavior from sparse 
data and enables real-time inference through parallelized CNN processing. Unlike traditional models 
that treat collaborative and content-based filtering separately, CBCF-CNN creates a unified 
representation that dynamically adapts to user preferences while promoting recommendation diversity 
and relevance. The model's ability to scale with massive datasets while maintaining low latency and 
high accuracy makes it suitable for deployment in large-scale platforms, offering a significant 
advancement over existing recommendation techniques.