Design of hybrid movie recommender system based on enhanced collaborative and content models with feature engineering

Abstract

A Recommendation System (RS) is a system that filters the newlineinformation and helps the users to choose the corresponding target from the newlinehuge amount of information available in online. The system recommends useful newlineand satisfactory products such as books, music, jokes, and movies for targeting newlineusers based on their interest. The collaborative and content-based filtering are newlinethe different systems often used towards designing the recommender system, newlinewhich predicts the recommended items based upon the user preferences. newlineHowever, the RS provides poor performance for scalability, data sparsity and newlinecold start. Moreover, the hybrid recommender system has combined both the newlinetechniques in multiple ways to overcome the shortcomings and recommend the newlinefinal outcomes. Hence, this research work focus to develop a new hybrid newlinerecommender system by combining the collaborative and content based newlinefiltering for better recommendation list based on user preference or interest. newlineInitially, the collaborative process takes place, in which the recommendations newlinebased on enhanced Balanced Iterative Reducing and Clustering using newlineHierarchies (BIRCH) clustering and Gradient Boost Tree (GBT) ensemble newlinemodel. newlineThe content based filtering is performed by Topic Modelling using newlineLatent Dirichlet Allocation Model. Both Collaborative Filtering (CF) and newlineContent Based Filtering (CBF) give top N recommendations to the user, the newlinefinal recommendation is made by combining these two recommendation newlinesystems. In Hybrid Recommendation System (HRS), scoring, item weight and newlineranking have created using feature engineering technique. The scoring is newlineperformed by the recommendation provided by the two recommender systems. newlineThe Item weight is generated from the popularity of the item among the users newlinein the dataset. newline