Recommendation System for Big Data Software Using Popularity Model and Collaborative Filtering

A recommender system is a model which has the ability to predict the list of items according to the user's preferences or ratings. Recommender systems have huge areas of application ranging from music, books, movies, search queries, and social sites to news. A recommender system filters information into various categories which may be personalized or non-personalized. Recommendation systems are very useful for the user in discovering those items which they may not have found else ways. These above-mentioned systems are quite similar to search queries except that search queries need to be explicitly provided by the user but recommender system may do it implicitly without the user having prior knowledge about its working. There are many algorithms used in recommendation systems such as restricted Boltzmann machines (RBM), slope one, singular value decomposition, alternating least squares. This paper presents an in-depth study and analysis of the following two algorithms: Review-based popularity model and collaborative filtering model. Popularity model is a straightforward model based on the popularity and rating of an item given to it by the user. Collaborative filtering model is a very efficient and popular model and is used by many popular organizations like Netflix, Amazon, and Facebook.