A Scalable Recommendation Engine for New Users and Items

TL;DR

This paper presents a scalable recommendation system that effectively addresses cold-start, preference learning, and scalability challenges by integrating collaborative filtering with multi-armed bandits and attributes, demonstrating significant improvements in real and synthetic data.

Contribution

It introduces the CFB-A model, a novel combination of collaborative filtering, multi-armed bandits, and attributes, to comprehensively solve key recommendation challenges.

Findings

Significant improvement in cumulative rewards over baseline methods.

Effective handling of cold-start and preference learning.

Validated through offline, synthetic, and online experiments.

Abstract

In many digital contexts such as online news and e-tailing with many new users and items, recommendation systems face several challenges: i) how to make initial recommendations to users with little or no response history (i.e., cold-start problem), ii) how to learn user preferences on items (test and learn), and iii) how to scale across many users and items with myriad demographics and attributes. While many recommendation systems accommodate aspects of these challenges, few if any address all. This paper introduces a Collaborative Filtering (CF) Multi-armed Bandit (B) with Attributes (A) recommendation system (CFB-A) to jointly accommodate all of these considerations. Empirical applications including an offline test on MovieLens data, synthetic data simulations, and an online grocery experiment indicate the CFB-A leads to substantial improvement on cumulative average rewards (e.g., total money or time spent, clicks, purchased quantities, average ratings, etc.) relative to the most powerful extant baseline methods.