Accurate and Diverse Recommendations via Propensity-Weighted Linear Autoencoders

TL;DR

This paper introduces a novel propensity scoring method using a sigmoid function on log-frequency to improve recommendation diversity in MNAR data, integrated into a linear autoencoder, achieving better diversity without losing accuracy.

Contribution

It redefines propensity scores with a sigmoid-log approach and incorporates them into a linear autoencoder to enhance recommendation diversity.

Findings

Significantly improves recommendation diversity.

Maintains recommendation accuracy.

Reduces bias towards popular items.

Abstract

In real-world recommender systems, user-item interactions are Missing Not At Random (MNAR), as interactions with popular items are more frequently observed than those with less popular ones. Missing observations shift recommendations toward frequently interacted items, which reduces the diversity of the recommendation list. To alleviate this problem, Inverse Propensity Scoring (IPS) is widely used and commonly models propensities based on a power-law function of item interaction frequency. However, we found that such power-law-based correction overly penalizes popular items and harms their recommendation performance. We address this issue by redefining the propensity score to allow broader item recommendation without excessively penalizing popular items. The proposed score is formulated by applying a sigmoid function to the logarithm of the item observation frequency, maintaining the simplicity of power-law scoring while allowing for more flexible adjustment. Furthermore, we incorporate the redefined propensity score into a linear autoencoder model, which tends to favor popular items, and evaluate its effectiveness. Experimental results revealed that our method substantially improves the diversity of items in the recommendation list without sacrificing recommendation accuracy.