Synthetic Data and Simulators for Recommendation Systems: Current State and Future Directions

Abstract

Synthetic data and simulators have the potential to markedly improve the performance and robustness of recommendation systems.These approaches have already had a beneficial impact in other machine-learning driven fields. We identify and discuss a key trade-off between data fidelity and privacy in the past work on synthetic data and simulators for recommendation systems. For the important use case of predicting algorithm rankings on real data from synthetic data, we provide motivation and current successes versus limitations. Finally we outline a number of exciting future directions for recommendation systems that we believe deserve further attention and work, including mixing real and synthetic data, feedback in dataset generation, robust simulations, and privacy-preserving methods.

Multi-Resolution Diffusion for Privacy-Sensitive Recommender Systems

“…The need for fully synthetic copies of datasets has been driven by the need of companies to stay compliant with data protection regulations [34], such as the GDPR (General Data Protection Regulation) [35]. Many works have been proposed to create data generators whose objective consists of synthesizing tabular datasets which hold the same statistical properties as the original, but are not identical copies [36]- [38].…”