Multifaceted Recommender System

Abstract

The current era of digitisation has resulted in the generation of a tremendous volume of newlinedigital content in almost every industry. The entertainment industry, especially movies, is no newlineexception. The introduction of OTT platforms and the availability of huge amounts of movie newlinedata online has raised the concern of recommending the right selection of movies as per mood newlineand requirements. So, there is a need for an efficient and personalised movie recommendation newlinesystem considering the multiple facets of users and movies. The movie recommender system newline(RecSys) is an information filtering system that helps viewers find the best movies. newlineRecSys identifies the viewer s interest and make movie recommendations. The RecSys newlinealgorithm must be able to predict ratings and suggest relevant recommendations to existing and newlinenew users. There are multiple challenges faced by traditional RecSys, which may affect the newlinequality of recommendations. newlineThere are two types of recommender systems: content-based and collaborative filteringbased newlinerecommender systems. The content-based recommender system does not require the newlineviewer s evaluation of movies for recommendation. Instead, the movie similarity is calculated newlinebased on movie features to make the recommendation. Collaborative filtering uses the movie newlineratings given by users to generate recommendations. newlineThis research introduces a novel multifaceted movie recommender system (MFRISE) for newlinerecommending movies by considering multiple facets of data and processing. The augmented newlinedataset is constructed by extracting features from the MovieLens dataset and other sources on newlinethe web, which offers multiple facets of data to study and evaluate the proposed system. These newlinefeatures are used to construct movie and user profiles to calculate their similarity. newlineThe novel MFRISE is designed using a hybrid recommendation approach to combine the newlineadvantages of both content-based and collaborative filtering. The accuracy of the proposed Hybrid RecSys algorithm is benchmarked with existing algorithms.