Addressing coldstart problem in recommender systems

Abstract

In the following three chapters of my thesis, I have applied several methods to solve the coldstart problem in Recommender Systems. The coldstart problem is the situation where a user or item is new to a website and recommendations need to be given to the new user or the new item needs to be recommended. The framework proposed in my work uses user's demographic information and the genre of movies for creating the model. The chapters propose a framework, parallelize it and also improve predictions and allows speedup by using different techniques to meet the given goal (alleviate the coldstart problem).