Collaborative Filtering under Model Uncertainty

TL;DR

This paper investigates how different parameter choices in a recommender system model affect item availability and ambiguity, revealing that most variations produce similar results with occasional significant differences.

Contribution

It applies the concept of predictive multiplicity to analyze model uncertainty in recommender systems, highlighting the impact of parameter variation on outcomes.

Findings

Most models yield similar availability results

Significant differences occur in some parameter settings

Model uncertainty can influence recommendation ambiguity

Abstract

In their work, Dean, Rich, and Recht create a model to research recourse and availability of items in a recommender system. We used the definition of predictive multiplicity by Marx, Pin Calmon, and Ustun to examine different variations of this model, using different values for two model parameters. Pairwise comparison of their models show, that most of these models produce very similar results in terms of discrepancy and ambiguity for the availability and only in some cases the availability sets differ significantly.