Improving incremental recommenders with online bagging

TL;DR

This paper explores the application of online bagging to incremental recommender systems, demonstrating significant accuracy improvements with minimal computational cost in data stream scenarios.

Contribution

It introduces the use of online bagging with incremental matrix factorization for top-N recommendations, a novel approach in data stream recommender systems.

Findings

Online bagging improves accuracy by up to 35%.

The method incurs small computational overhead.

Effective for positive-only binary ratings in data streams.

Abstract

Online recommender systems often deal with continuous, potentially fast and unbounded flows of data. Ensemble methods for recommender systems have been used in the past in batch algorithms, however they have never been studied with incremental algorithms that learn from data streams. We evaluate online bagging with an incremental matrix factorization algorithm for top-N recommendation with positive-only -- binary -- ratings. Our results show that online bagging is able to improve accuracy up to 35% over the baseline, with small computational overhead.