A Study on Hybrid Recommender System with Deep Learning and Deployment in Big Data

Abstract

A recommender system is a filtering tool that provides customized suggestions of products to users by using various techniques. A hybrid recommender system combines the approaches of two or more techniques, primarily focusing on Content-based recommender systems and Collaborative recommender systems, thus proving better suggestions by exploiting the advantages of both the strategies. This paper provides a detailed review of the various ways to implement a Hybrid Recommender System, enabling the reader to get a bird’s eye view on the concept. The study allows the user to understand the relationship between data characteristics and the relative performance of various approaches to a Hybrid Recommender System. The flow of the research  starts with a description of the initial models of Recommender Systems, followed by a study on the application of Machine Learning and Deep  Learning algorithms to the Hybrid Recommenders. The contributions to the study are dichotomous: 1) to identify and discuss the various Techniques for Hybrid Recommendation Systems via a Systematic literature review and 2) to compare the various techniques to get a clear view of the efficiency and the accuracy of each technique. At the end of the study, the reader understands the development flow of Hybrid Recommender Systems and gets a clear idea of how various Recommender Systems work. It is observed that the technique which incorporates Deep Learning provides better results.