Predicting Item Popularity: Analysing Local Clustering Behaviour of Users

TL;DR

This paper introduces a novel method for predicting the future popularity of new items in rating networks by analyzing the local clustering behavior of the first user who rates the item, achieving over 65% success rate.

Contribution

It proposes a new bipartite clustering coefficient and demonstrates its effectiveness in predicting item popularity in MovieLens and Digg networks.

Findings

Over 65% success rate in MovieLens

Over 50% success rate in Digg

Improves upon existing prediction methods

Abstract

Predicting the popularity of items in rating networks is an interesting but challenging problem. This is especially so when an item has first appeared and has received very few ratings. In this paper, we propose a novel approach to predicting the future popularity of new items in rating networks, defining a new bipartite clustering coefficient to predict the popularity of movies and stories in the MovieLens and Digg networks respectively. We show that the clustering behaviour of the first user who rates a new item gives insight into the future popularity of that item. Our method predicts, with a success rate of over 65% for the MovieLens network and over 50% for the Digg network, the future popularity of an item. This is a major improvement on current results.