Cold-start recommendations in Collective Matrix Factorization

TL;DR

This paper investigates cold-start recommendation capabilities of collective matrix factorization, proposing a faster prediction method that enhances cold-start performance over non-personalized methods, especially for new users.

Contribution

It introduces a new, faster formulation for cold-start predictions in collective matrix factorization models, improving their real-time applicability and recommendation quality for new users.

Findings

Cold-start recommendations outperform non-personalized methods.

Predictions for new users are more reliable than for new items.

The new formulation improves cold-start results at the cost of warm-start performance.

Abstract

This work explores the ability of collective matrix factorization models in recommender systems to make predictions about users and items for which there is side information available but no feedback or interactions data, and proposes a new formulation with a faster cold-start prediction formula that can be used in real-time systems. While these cold-start recommendations are not as good as warm-start ones, they were found to be of better quality than non-personalized recommendations, and predictions about new users were found to be more reliable than those about new items. The formulation proposed here resulted in improved cold-start recommendations in many scenarios, at the expense of worse warm-start ones.