Genre-Aware Hybrid Recommendation Model for Improving Movie Discovery Using Metadata, Viewer Ratings ,and Plot Summaries

Abstract

This paper presents a Genre-Aware Hybrid Deep Learning Recommendation Model designed to enhance movie discovery by integrating heterogeneous information sources, including metadata, viewer ratings, and plot summaries. Unlike traditional collaborative or content-based systems that rely on a single data modality, the proposed approach leverages genre-aware embeddings to capture contextual similarities between movies while incorporating user preference patterns from historical ratings. A deep neural architecture combines convolutional and recurrent layers to process textual plot summaries, while dense layers handle structured metadata and rating data. The fusion mechanism enables a unified latent representation, improving recommendation accuracy and diversity. Experimental evaluation on benchmark movie datasets demonstrates that the proposed model outperforms state-of-the-art baselines in precision, recall, and normalized discounted cumulative gain (). The genre-aware design also enhances cold-start performance by effectively utilizing textual and categorical features, thereby delivering more personalized and contextually relevant movie recommendations for diverse user profiles.