A Comparative Study of Collaborative Filtering Algorithms

TL;DR

This paper compares various collaborative filtering algorithms across different conditions to identify which techniques perform best and under what circumstances, aiding both research and practical deployment.

Contribution

It provides a comprehensive comparative analysis of classic and recent collaborative filtering algorithms under diverse experimental settings.

Findings

Certain algorithms excel with dense data

Some methods perform better with sparse data

Trade-offs between accuracy and computational complexity

Abstract

Collaborative filtering is a rapidly advancing research area. Every year several new techniques are proposed and yet it is not clear which of the techniques work best and under what conditions. In this paper we conduct a study comparing several collaborative filtering techniques -- both classic and recent state-of-the-art -- in a variety of experimental contexts. Specifically, we report conclusions controlling for number of items, number of users, sparsity level, performance criteria, and computational complexity. Our conclusions identify what algorithms work well and in what conditions, and contribute to both industrial deployment collaborative filtering algorithms and to the research community.