Online Partitioning of Large Graphs for Improving Scalability in Recommender Systems

Abstract

All around the world people are using social media to create large online social networks, and the sizes of these networks are continuously increasing. It is estimated that by the year 2018, 2.44 billion people will be using social networks. The growing desire and demand of social networks are not just removing our communication barrier, and it is also changing the way we think and do things. The people we ‘trust’ or ‘are close to’ in our social network continuously shape our views about different things. This trust-based relationship can be harnessed to recommend items to users. In this paper, we aim to tackle the problem of scalability in online recommender systems using an online graph partitioning algorithm.