CineFinder: A Movie Recommendation System Using Visual and Textual Deep Features

Mehmet Tuğrul Sarıçiçek, Rukiye Orman, Murat Dener, Harun Kınacı

2025

Abstract

In recent years, the increasing popularity of digital content platforms has highlighted the need for personalized recommendation systems, particularly in the entertainment industry. Traditional recommendation systems often suffer from limitations such as the "cold start" problem and inadequate personalization due to their reliance on limited user data. To address these challenges, this study proposes CineFinder. This hybrid feature-based movie recommendation system integrates both visual and textual deep features using multiple state-of-the-art pre-trained models. CineFinder extracts visual features from movie posters and backdrops using pre-trained convolutional neural networks—namely VGG-16, ResNet-50, and MobileNet—and captures textual features from movie overviews using pre-trained transformer-based models such as BERT, RoBERTa, and SBERT. These extracted features are fused into a comprehensive hybrid feature vector and utilized for similarity-based recommendations via Cosine similarity, Euclidean distance, and Manhattan distance. The system's performance was evaluated on two datasets created by the authors: the TMDB Dataset, which provides general audience metrics, and the TMDBRatingsMatched Dataset, which incorporates user-specific rating data from MovieLens 20M. Experimental results demonstrate that the proposed approach generates accurate and relevant movie recommendations while mitigating the cold start problem. The findings highlight the effectiveness of integrating multimodal deep learning techniques and leveraging user-driven feedback to enhance recommendation accuracy.

Paper Citation

in Harvard Style

Sarıçiçek M., Orman R., Dener M. and Kınacı H. (2025). CineFinder: A Movie Recommendation System Using Visual and Textual Deep Features. In Proceedings of the 2nd International Conference on Advances in Electrical, Electronics, Energy, and Computer Sciences - Volume 1: ICEEECS; ISBN 978-989-758-783-2, SciTePress, pages 120-131. DOI: 10.5220/0014385700004848

in Bibtex Style

@conference{iceeecs25,
author={Mehmet Tuğrul Sarıçiçek and Rukiye Orman and Murat Dener and Harun Kınacı},
title={CineFinder: A Movie Recommendation System Using Visual and Textual Deep Features},
booktitle={Proceedings of the 2nd International Conference on Advances in Electrical, Electronics, Energy, and Computer Sciences - Volume 1: ICEEECS},
year={2025},
pages={120-131},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0014385700004848},
isbn={978-989-758-783-2},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Advances in Electrical, Electronics, Energy, and Computer Sciences - Volume 1: ICEEECS
TI - CineFinder: A Movie Recommendation System Using Visual and Textual Deep Features
SN - 978-989-758-783-2
AU - Sarıçiçek M.
AU - Orman R.
AU - Dener M.
AU - Kınacı H.
PY - 2025
SP - 120
EP - 131
DO - 10.5220/0014385700004848
PB - SciTePress