Modeling User Preferences as Distributions for Optimal Transport-Based Cross-Domain Recommendation under Non-Overlapping Settings

TL;DR

This paper introduces DUP-OT, a novel non-overlapping cross-domain recommender system that models user preferences as Gaussian mixtures and uses optimal transport for effective knowledge transfer, improving cold-start recommendations.

Contribution

DUP-OT is the first framework to model user preferences as Gaussian mixtures and align them across domains using optimal transport without overlapping users or items.

Findings

DUP-OT outperforms single-domain baselines in RMSE.

DUP-OT achieves lower RMSE than TDAR in non-overlapping settings.

Effective in cold-start user preference transfer.

Abstract

Cross-domain recommender (CDR) systems aim to transfer knowledge from data-rich domains to data-sparse ones, alleviating sparsity and cold-start issues present in conventional single-domain recommenders. However, many CDR approaches rely on overlapping users or items to establish explicit cross-domain connections, which is unrealistic in practice. Moreover, most methods represent user preferences as fixed discrete vectors, limiting their ability to capture the fine-grained and multi-aspect nature of user interests. To address these limitations, we propose DUP-OT (Distributional User Preferences with Optimal Transport), a novel framework for non-overlapping CDR. DUP-OT consists of three stages: (1) a shared preprocessing module that extracts review-based embeddings using a unified sentence encoder and autoencoder; (2) a user preference modeling module that represents each user's interests as a Gaussian Mixture Model (GMM) over item embeddings; and (3) an optimal-transport-based alignment module that matches Gaussian components across domains, enabling effective preference transfer for target-domain rating prediction. Experiments on Amazon Review datasets show that DUP-OT outperforms single-domain baselines even without source-domain data, and achieves lower RMSE than the cross-domain baseline TDAR under strictly non-overlapping training settings, demonstrating its effectiveness in reducing large prediction errors for cold-start users. The implementation is available at https://github.com/XiaoZY2000/dup-ot.