Divergence Meets Consensus: A Multi-Source Negative Sampling Framework for Sequential Recommendation

2026-05-19

AI Summary

Existing self-guided hard negative sampling methods often suffer from suboptimal local convergence, insufficient diversity in negative samples, and high computational overhead. To address these limitations, this work proposes the MDCNS framework, inspired by Vygotsky’s “Zone of Proximal Development” theory. MDCNS introduces a novel teacher–peer–self triadic structure that incorporates multi-source external signals—used for the first time to break self-reinforcing feedback loops—and employs prediction-discrepancy-driven reranking to enhance sample diversity. A KL divergence–based consensus distillation mechanism efficiently integrates knowledge from multiple sources. Extensive experiments demonstrate that MDCNS consistently outperforms state-of-the-art methods across six real-world datasets and five backbone architectures, achieving superior performance and strong generalization capability.

Abstract

Negative sampling is significant for training sequential recommendation models under implicit feedback. The predominant strategy, self-guided hard negative sampling, selects negatives based on the model's current state but suffers from three limitations: (1) the coupling between sampling and model updates triggers a vicious cycle that drives the model into local optima; (2) relying on current model parameters narrows sampling to a small region of the item space, reducing diversity and harming generalization; (3) identifying a hard negative requires scoring the entire candidate pool, causing substantial computational overhead with minimal information gain. To address these challenges, we propose MDCNS (Multi-source Divergence-Consensus for Negative Sampling), a novel "Teacher-Peer-Self" framework inspired by Vygotsky's Zone of Proximal Development (ZPD) theory. The proposed method comprises three components, including multi-source scoring, divergence re-ranking, and consensus distillation. Firstly, multi-source scoring incorporates peer and ensemble teacher models to inject external negative signals and break the self-reinforcement loop. Then, divergence re-ranking exploits prediction discrepancy between self and peer models to enhance sampling diversity. Finally, consensus distillation aligns the self model with the teacher via KL divergence, simultaneously improving computational cost utilization. Extensive experiments on six real-world datasets and five backbone models show that MDCNS consistently outperforms state-of-the-art negative sampling methods, demonstrating strong effectiveness and generalization.

Problem

• negative sampling
• sequential recommendation
• hard negatives
• model coupling
• sampling diversity

Innovation

• negative sampling
• sequential recommendation
• multi-source learning
• divergence-consensus
• knowledge distillation