Improving the accuracy of movie recommendation using collaborative filtering

Movie Recommendation Systems (MRS) play a pivotal role in delivering personalized cinematic content by analyzing users' preferences, interactions, and behavioral patterns. However, one of the persistent challenges faced by such systems is the cold start problem, which occurs due to insufficient initial user data, thereby constraining the system's ability to produce accurate recommendations for new users. This study aims to mitigate the cold start problem in MRS by introducing a hybrid recommendation framework that incorporates emotion-to-genre translation into the recommendation process. The proposed method integrates content-based and collaborative filtering techniques with an emotion-driven mapping model to enhance personalization and predictive accuracy. Experimental evaluations demonstrate that the proposed framework achieves superior performance in terms of accuracy compared to state-of-the-art methods. In particular, the four-model ensemble configuration exhibited higher precision, recall, and F۱-score than both DNN-based and ILDNet-based baselines.

کلیدواژه ها:

Movie recommendation system ، Cold start ، User emotion ، Collaborative filtering ، Content-based filtering ، Ensemble learning