Enhancing item based collaborative filtering

Abstract

Recommendations play an important role in this contemporary era. Most use online platforms like Amazon, Net- flix, etc. to buy daily necessities and entertainment. Moreover, the referral system helps both consumers and customers regardless of age group and financial background. There are mainly three types of referral systems namely Content-Based, Collaborative and Hybrid. Collaborative systems have many problems and some of them are solved to some extent, but the cold start problem is hardly solved by the researcher. Traditional recommendation algorithms such as matrix factorization and collaborative filtering perform poorly when they lack information on the interaction between the user and the product, known as the user cold-start problem, which can cause reduce the revenue of the e-commerce platform. Cold - star problem is divided into two types User cold star and item cold star. In this search, we have selected cross- domain to solve item coldstar and it works fine for cold non-star issues as well. We choose the Amazon dataset for deployment where one is considered the source domain and the other as the target domain.