Challenges and Solutions in Recommender Systems

Abstract

Recommender Systems play a huge part in most of our lives today. A large portion of today’s digital customers rely on such programs to shape their usage of online markets. The main objective of such systems is to build relationships between the products and its users and to help them make the best decisions depending on their needs. There are 4 main types of Recommender Systems that follow different methods in order to satisfy user preferences by filtering through data in an efficient manner. Content-based filtering systems suggest items based on attributes they share with other similar items. Collaborative filtering systems analyze the past behaviour of the customer and recommend items they might find interesting. Demographic filtering uses pre-existing and already compiled data about the behaviour of the customer based on population statistics and uses it to create a list of recommendations and the Hybrid filtering system is a combination of all these systems. Essentially, recommender systems filter through large amounts of data to give the user personalized results. This paper explains and analyzes the Filtering Systems in depth and delves into the challenges these systems face in order to produce accurate suggestions.