Reducing offline evaluation bias of collaborative filtering algorithms

TL;DR

This paper introduces a weighted offline evaluation method to mitigate bias in assessing collaborative filtering recommendation algorithms, addressing the influence of prior system interactions on evaluation accuracy.

Contribution

It proposes a novel application of weighted offline evaluation specifically designed to reduce bias in collaborative filtering algorithm assessment.

Findings

Weighted evaluation reduces bias in offline assessments

Improved accuracy of recommendation performance measurement

Method applicable to large-scale online systems

Abstract

Recommendation systems have been integrated into the majority of large online systems to filter and rank information according to user profiles. It thus influences the way users interact with the system and, as a consequence, bias the evaluation of the performance of a recommendation algorithm computed using historical data (via offline evaluation). This paper presents a new application of a weighted offline evaluation to reduce this bias for collaborative filtering algorithms.