Efficient Cohort Nearest Neighbor (CNN) Technique to leverage Recommender Systems

The long tail of diverse consumption of resources online by the customers raise a challenge for the e-commerce websites and service providers. Recommender system offers a vigorous way to cope up with the aforementioned challenge. With the advent of the internet, these systems have achieved widespread success in e-commerce applications. The most widely used recommender system technique is collaborative filtering which guides users in a personalized way to suggest items from a large set of possible options. In this paper, we have proposed a cohort nearest neighbour (CNN) recommender system which relies on high cohort users to make predictions. In our work, we have used MovieLens dataset to perform our experiments and measurement of accuracy with Root Mean Square Error (RMSE) technique. This approach significantly improves the retention of recommender system by more than 1.09 times when the granularity of data space increases.