Score-based Generative Diffusion Models for Social Recommendations

2024-12-20

arXiv.org

AI Summary

To address degraded recommendation performance in real-world social networks—characterized by weak social homophily and strong noise—this paper proposes SGSR, a score-based generative diffusion model for social recommendation. SGSR is the first to introduce stochastic differential equation (SDE)-based diffusion modeling into this domain. It learns robust user social representations via score matching and jointly incorporates a curriculum learning strategy with a social–collaborative dual-domain self-supervised alignment mechanism, enhancing collaborative signal consistency without relying on strong homophily assumptions. Extensive experiments on multiple real-world datasets demonstrate that SGSR effectively suppresses redundant social noise, achieving state-of-the-art (SOTA) performance in both accuracy and recall. These results validate the efficacy and generalizability of generative modeling for social recommendation under low-homophily conditions.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Social recommender systems and personalization for the social Web
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

With the prevalence of social networks on online platforms, social recommendation has become a vital technique for enhancing personalized recommendations. The effectiveness of social recommendations largely relies on the social homophily assumption, which presumes that individuals with social connections often share similar preferences. However, this foundational premise has been recently challenged due to the inherent complexity and noise present in real-world social networks. In this paper, we tackle the low social homophily challenge from an innovative generative perspective, directly generating optimal user social representations that maximize consistency with collaborative signals. Specifically, we propose the Score-based Generative Model for Social Recommendation (SGSR), which effectively adapts the Stochastic Differential Equation (SDE)-based diffusion models for social recommendations. To better fit the recommendation context, SGSR employs a joint curriculum training strategy to mitigate challenges related to missing supervision signals and leverages self-supervised learning techniques to align knowledge across social and collaborative domains. Extensive experiments on real-world datasets demonstrate the effectiveness of our approach in filtering redundant social information and improving recommendation performance.

Problem

• Addressing low social homophily challenge in recommendation systems
• Generating optimal user representations aligning with collaborative signals
• Filtering redundant social information to improve recommendation performance

Innovation

• Score-based generative model for social recommendation
• Stochastic Differential Equation diffusion models adaptation
• Joint curriculum training with self-supervised learning