Recommending E-Commerce Products on Cold Start and Long Tail Using Transaction Data

Abstract

Recommender method plays a critical role in the discovery and display of interesting items from near-infinite inventories to potential users. Yet the recommender systems are being challenged by two problems. One is “how to address new users” and the other one is “how to amaze users”. The problem is known as a cold-start recommendation. In this paper, how the products are handled in the e-commerce retailers getting sold out soon, and products remain stagnant for a long duration isdiscussed. To overcome this problem, the selling of products to customers of the same kind based is tracked on the previous purchase by using clustering and classification algorithms. To recommend combo products to users and general recommendation based on user point of interest and product transaction details. This way the retailers are getting helped to gain some profit which is not sold for a long period.