A DEEP LEARNING-BASED SYSTEM FOR RECOMMENDING MOVIES BASED ON EMOTIONS

Abstract: In the era of intelligent entertainment, conventional movie recommendation systems oftenfail to capture the dynamic emotional preferences of users. This paper proposes a novel emotion-basedmovie recommendation system that utilizes facial expression recognition to detect a user's real-timeemotional state and generate personalized movie suggestions accordingly. Using convolutional neuralnetworks (CNNs) for facial emotion classification, the system maps detected emotions to appropriatemovie genres through a predefined emotion-genre mapping model. The proposed system enhancesuser satisfaction by adapting to momentary emotional contexts, unlike static, history-based methods.Experimental results demonstrate that the emotion-aware approach significantly improvesrecommendation relevance and user engagement. Future extensions may include multimodal emotiondetection and integration with user profiles for deeper personalization. Keywords—Convolutional Neural Networks (CNNs), VGG16, and ResNet, Movie Recommendation,Emotions.