Federated Recommender System with Data Valuation for E-commerce Platform

2025-09-14

AI Summary

In e-commerce federated recommendation, injecting noise into global data (e.g., user/item catalogs) undermines model personalization. To address this, we propose a semantics-aligned selective enhancement method for global data. Our approach features: (1) an RL-driven probabilistic estimator that dynamically selects high-quality global samples semantically consistent with local interaction patterns; (2) integration of a pre-trained graph encoder to extract structural features; and (3) a local validity predictor that guides federated collaborative training. The method preserves data privacy while effectively suppressing interference from irrelevant global patterns, thereby enhancing local model personalization. Evaluated on mainstream federated recommendation benchmarks, it achieves up to a 34.86% improvement in Recall@10—outperforming state-of-the-art methods. This work is the first to systematically demonstrate the efficacy and scalability of principled global-data valuation and controlled utilization in federated recommendation.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Search and Optimization: Local Search
• Data Mining & Knowledge Management: Recommender Systems

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Federated Learning (FL) is gaining prominence in machine learning as privacy concerns grow. This paradigm allows each client (e.g., an individual online store) to train a recommendation model locally while sharing only model updates, without exposing the raw interaction logs to a central server, thereby preserving privacy in a decentralized environment. Nonetheless, most existing FL-based recommender systems still rely solely on each client's private data, despite the abundance of publicly available datasets that could be leveraged to enrich local training; this potential remains largely underexplored. To this end, we consider a realistic scenario wherein a large shopping platform collaborates with multiple small online stores to build a global recommender system. The platform possesses global data, such as shareable user and item lists, while each store holds a portion of interaction data privately (or locally). Although integrating global data can help mitigate the limitations of sparse and biased clients' local data, it also introduces additional challenges: simply combining all global interactions can amplify noise and irrelevant patterns, worsening personalization and increasing computational costs. To address these challenges, we propose FedGDVE, which selectively augments each client's local graph with semantically aligned samples from the global dataset. FedGDVE employs: (i) a pre-trained graph encoder to extract global structural features, (ii) a local valid predictor to assess client-specific relevance, (iii) a reinforcement-learning-based probability estimator to filter and sample only the most pertinent global interactions. FedGDVE improves performance by up to 34.86% on recognized benchmarks in FL environments.

Problem

• Integrating global data into federated recommenders without noise amplification
• Addressing sparse and biased local data in decentralized recommendation systems
• Selectively augmenting client graphs with semantically aligned global samples

Innovation

• Selectively augments local graphs with global data
• Uses pre-trained graph encoder for structural features
• Employs reinforcement learning to filter relevant interactions