Stacking Based Meta Model for Movie Recommendation

Abstract

Entertainment industry in this internet era has constantly been taking huge interest in ensuring a tailored experience to each of its audience and human constantly searching for better choices. This paper proposes a Recommender model using collaborative filtering and machine learning based on the stacking model which is one of the ensemble learning method that can handle cold start and diversity problems thereby providing more reliable predictions. Stacking processes are essential in a movie recommendation system, which combines many base learners through a meta learner. User and item-based collaborative filtering are combined to identify the probability of items. These items are used to train the meta level base learner model to predict user ratings on new items and to provide final recommendations. The experimental result of the proposed model has been compared with that of the state of art models in terms of MSE, MAE and RMSE.