Master's Thesis

Video Recommendations Based on Visual Features Extracted with Deep Learning

When a movie is uploaded to a movie Recommender System (e.g., YouTube), the system can exploit various forms of descriptive features (e.g., tags and genre) in order to generate personalized recommendation for users. However, there are situations where the descriptive features are missing or very limited and the system may fail to include such a movie in the recommendation list, known as Cold-start problem. This thesis investigates recommendation based on a novel form of content features, extracted from movies, in order to generate recommendation for users. Such features represent the visual aspects of movies, based on Deep Learning models, and hence, do not require any human annotation when extracted. The proposed technique has been evaluated in both offline and online evaluations using a large dataset of movies. The online evaluation has been carried out in a evaluation framework developed for this thesis. Results from the offline and online evaluation (N=150) show that automatically extracted visual features can mitigate the cold-start problem by generating recommendation with a superior quality compared to different baselines, including recommendation based on human-annotated features. The results also point to subtitles as a high-quality future source of automatically extracted features.

SAMVISE - Demo of framework for evaluating movie recommendation systems, available at: https://samvise.herokuapp.com/

Datasets - The datasets related to the thesis are all published and openly available at: https://zenodo.org/record/4889729