A Comprehensive Study of Hybrid Recommendation Systems for E-Commerce Applications

Abstract

Recommender Systems (RS) are playing a significant role to improve the business growth of web-based applications around the world.  It is moderately a new area of research in Artificial Intelligence and the fundamental idea is to build a connection between the items, users and make a choice to choose the most appropriate item to a particular user. The current research of RS can be mainly on content-based, collaborative-based, and hybrid recommendation approaches. The Content-based approach utilizes the profiles of the users and product descriptions for the recommendation of items. Collaborative techniques utilized the behavior of the users and preferences by the user in the past unlike the user’s personal information and product description. Hybrid recommendation systems include the strategies of content-based and collaborative based on effective recommendations. There exist many works in the literature, in which the recommendation of books, articles, and news, etc. are utilizing either content-based or collaborative based techniques and very few works have been done on hybrid recommendation methods. In this paper, we addressed a comprehensive analysis of hybrid recommendation systems used in numerous applications such as e-commerce, e-learning, movies, etc. First, we presented the essential terminologies and the fundamental concepts of RS.  Second, we have given a brief review of hybrid recommendation techniques proposed in recent years along with the implementation process. In addition to this, we also presented the evaluation metrics for calculating the performance of a recommendation model.