FedDCSR: Federated Cross-domain Sequential Recommendation via Disentangled Representation Learning

TL;DR

This work tackles privacy-preserving cross-domain sequential recommendation by enabling federated training across domains. It introduces SRD to disentangle domain-shared and domain-exclusive sequence representations and CIM to enhance domain-exclusive features via intra-domain infomax, all built atop a variational graph self-attention encoder (VGSE). The approach is validated on three Amazon CSR scenarios, where FedDCSR consistently outperforms strong baselines and ablations confirm the contributions of SRD and CIM for improved cross-domain personalization under privacy constraints. The results underscore the practical potential of disentangled representation learning in federated settings for privacy-preserving, multi-domain recommender systems.

Abstract

Cross-domain Sequential Recommendation (CSR) which leverages user sequence data from multiple domains has received extensive attention in recent years. However, the existing CSR methods require sharing origin user data across domains, which violates the General Data Protection Regulation (GDPR). Thus, it is necessary to combine federated learning (FL) and CSR to fully utilize knowledge from different domains while preserving data privacy. Nonetheless, the sequence feature heterogeneity across different domains significantly impacts the overall performance of FL. In this paper, we propose FedDCSR, a novel federated cross-domain sequential recommendation framework via disentangled representation learning. Specifically, to address the sequence feature heterogeneity across domains, we introduce an approach called inter-intra domain sequence representation disentanglement (SRD) to disentangle the user sequence features into domain-shared and domain-exclusive features. In addition, we design an intra domain contrastive infomax (CIM) strategy to learn richer domain-exclusive features of users by performing data augmentation on user sequences. Extensive experiments on three real-world scenarios demonstrate that FedDCSR achieves significant improvements over existing baselines.

Figures (7)

• Figure 1: Sequence feature heterogeneity across domains in the FedCSR scenario.
• Figure 2: An overview of FedDCSR.
• Figure 3: The architecture of the variational graph self-attention encoder.
• Figure 4: The graphical model illustrating the relationship between $\boldsymbol{Z}^{\mathrm{s}}_k$, $\boldsymbol{Z}^{\mathrm{e}}_k$, $\boldsymbol{Z}^{\text{g}}_k$, $\boldsymbol{S}_k$ and the dummy variable $\boldsymbol{S}^{\mathrm{g}}$, which follows the global squence distribution.
• Figure 5: FedDCSR