Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Personalized Privacy: User-Governed Data Contribution for Federated Recommendation

Liang Qu, Wei Yuan, Ruiqi Zheng, Lizhen Cui, Yuhui Shi, Hongzhi Yin

TL;DR

This work tackles privacy-sensitive personalization in federated recommender systems by allowing users to govern the amount of data they share. It introduces CDCGNNFed, a cloud-device collaborative graph neural network that trains user-centric ego graphs locally while building high-order server graphs from user-contributed data via a graph-mending strategy, and aligns local and global views through device-server contrastive learning. Empirical results on Gowalla and Yelp2018 show the approach outperforms baselines in partial-upload scenarios, with ablations demonstrating the necessity of graph mending and contrastive learning for capturing higher-order structure and robust embeddings. The framework offers a practical path toward flexible privacy budgets in FedRecs, enabling improved recommendations without forcing all users to surrender data, and highlights the value of data-sharing still allowed by users in boosting model performance.

Abstract

Federated recommender systems (FedRecs) have gained significant attention for their potential to protect user's privacy by keeping user privacy data locally and only communicating model parameters/gradients to the server. Nevertheless, the currently existing architecture of FedRecs assumes that all users have the same 0-privacy budget, i.e., they do not upload any data to the server, thus overlooking those users who are less concerned about privacy and are willing to upload data to get a better recommendation service. To bridge this gap, this paper explores a user-governed data contribution federated recommendation architecture where users are free to take control of whether they share data and the proportion of data they share to the server. To this end, this paper presents a cloud-device collaborative graph neural network federated recommendation model, named CDCGNNFed. It trains user-centric ego graphs locally, and high-order graphs based on user-shared data in the server in a collaborative manner via contrastive learning. Furthermore, a graph mending strategy is utilized to predict missing links in the graph on the server, thus leveraging the capabilities of graph neural networks over high-order graphs. Extensive experiments were conducted on two public datasets, and the results demonstrate the effectiveness of the proposed method.

Towards Personalized Privacy: User-Governed Data Contribution for Federated Recommendation

TL;DR

This work tackles privacy-sensitive personalization in federated recommender systems by allowing users to govern the amount of data they share. It introduces CDCGNNFed, a cloud-device collaborative graph neural network that trains user-centric ego graphs locally while building high-order server graphs from user-contributed data via a graph-mending strategy, and aligns local and global views through device-server contrastive learning. Empirical results on Gowalla and Yelp2018 show the approach outperforms baselines in partial-upload scenarios, with ablations demonstrating the necessity of graph mending and contrastive learning for capturing higher-order structure and robust embeddings. The framework offers a practical path toward flexible privacy budgets in FedRecs, enabling improved recommendations without forcing all users to surrender data, and highlights the value of data-sharing still allowed by users in boosting model performance.

Abstract

Federated recommender systems (FedRecs) have gained significant attention for their potential to protect user's privacy by keeping user privacy data locally and only communicating model parameters/gradients to the server. Nevertheless, the currently existing architecture of FedRecs assumes that all users have the same 0-privacy budget, i.e., they do not upload any data to the server, thus overlooking those users who are less concerned about privacy and are willing to upload data to get a better recommendation service. To bridge this gap, this paper explores a user-governed data contribution federated recommendation architecture where users are free to take control of whether they share data and the proportion of data they share to the server. To this end, this paper presents a cloud-device collaborative graph neural network federated recommendation model, named CDCGNNFed. It trains user-centric ego graphs locally, and high-order graphs based on user-shared data in the server in a collaborative manner via contrastive learning. Furthermore, a graph mending strategy is utilized to predict missing links in the graph on the server, thus leveraging the capabilities of graph neural networks over high-order graphs. Extensive experiments were conducted on two public datasets, and the results demonstrate the effectiveness of the proposed method.
Paper Structure (23 sections, 8 equations, 3 figures, 5 tables)

This paper contains 23 sections, 8 equations, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Centralized Recommendation
  4. Federated Recommendation
  5. Proposed Method
  6. Problem formulation
  7. CDCGNNFed
  8. Graph mending
  9. Embedding inference
  10. Device-server constrastive learning
  11. Device and server model training
  12. Global model aggregation
  13. Experiments
  14. Settings
  15. Datasets
  16. ...and 8 more sections

Figures (3)

  • Figure 1: (a) 0 privacy budget federated recommender systems that users do not upload any data to the server. (b) Personalized privacy budget federated recommender systems that users are free to take control of whether they share data and the proportion of data they share to the server.
  • Figure 2: The architecture of the proposed method.
  • Figure 3: The performance of model with different hyper-parameter settings on Gowalla dataset. (a) Threshold $t$; (b) Temperature $\tau$; (c) The number of devices $|\mathcal{U}_{s}|$; (d) The number of layers $l$.