GCIB: Graph Contrastive Information Bottleneck for Multi-Behavior Recommendation

2026-05-25

AI Summary

This work addresses the challenges in multi-behavior recommendation, where auxiliary behavior graphs often contain task-irrelevant noise and the supervision from target behaviors is sparse, hindering effective collaborative signal extraction. To tackle these issues, the authors propose a dual-level optimization framework: at the structural level, they introduce Graph Information Bottleneck (GIB) to perform task-oriented denoising of auxiliary graphs; at the feature level, they design a cross-behavior Graph Contrastive Learning (GCL) mechanism to enhance target-behavior representations. This study presents the first integration of GIB and GCL, enabling noise-robust semantic transfer and target-aware representation learning. Extensive experiments on multiple benchmark datasets demonstrate that the proposed method significantly outperforms state-of-the-art models, validating its superior denoising capability and representational effectiveness.

Abstract

With the rapid emergence of multi-behavior learning in recommender systems, leveraging auxiliary user behaviors has proven effective for mitigating target-behavior data sparsity. Yet auxiliary behavior graphs frequently contain noisy or irrelevant interactions that do not align with the target task, impeding the learning of accurate user and item embeddings. Moreover, the scarcity of direct supervision from the target behavior complicates the extraction of informative collaborative signals. In this paper, we introduce GCIB (Graph Contrastive Information Bottleneck), a novel framework that denoises auxiliary behavior information and enriches target behavior representations at both the structural and feature levels. At the structural level, GCIB employs a Graph Information Bottleneck (GIB) objective to maximize mutual information between the denoised auxiliary graph and the target-behavior graph while minimizing mutual information with the original auxiliary graph. This formulation preserves task-relevant structural patterns and suppresses spurious interactions. At the feature level, we propose a cross-behavior Graph Contrastive Learning (GCL) scheme in which denoised auxiliary features and target-behavior features serve as complementary views for both users and items. By contrasting these views, GCIB enriches sparse target-behavior representations with semantics distilled from auxiliary behaviors. Extensive experiments demonstrate that GCIB outperforms state-of-the-art baselines, highlighting its ability to learn noise-resilient and target-aware representations for multi-behavior recommendation.

Problem

• multi-behavior recommendation
• auxiliary behavior noise
• data sparsity
• collaborative signals
• representation learning

Innovation

• Graph Information Bottleneck
• Graph Contrastive Learning
• Multi-Behavior Recommendation
• Denoising
• Mutual Information