Contrastive Learning with Bidirectional Transformers for Sequential Recommendation

TL;DR

This paper introduces CBiT, a bidirectional Transformer-based contrastive learning framework for sequential recommendation, improving upon unidirectional models by better capturing user behavior sequences and employing multi-pair contrastive learning.

Contribution

The paper proposes a novel bidirectional Transformer framework with multi-pair contrastive learning and dynamic loss reweighting for enhanced sequential recommendation.

Findings

CBiT outperforms state-of-the-art models on benchmark datasets.

Bidirectional modeling captures user behavior more effectively.

Multi-pair contrastive learning improves sample quality and model robustness.

Abstract

Contrastive learning with Transformer-based sequence encoder has gained predominance for sequential recommendation. It maximizes the agreements between paired sequence augmentations that share similar semantics. However, existing contrastive learning approaches in sequential recommendation mainly center upon left-to-right unidirectional Transformers as base encoders, which are suboptimal for sequential recommendation because user behaviors may not be a rigid left-to-right sequence. To tackle that, we propose a novel framework named \textbf{C}ontrastive learning with \textbf{Bi}directional \textbf{T}ransformers for sequential recommendation (\textbf{CBiT}). Specifically, we first apply the slide window technique for long user sequences in bidirectional Transformers, which allows for a more fine-grained division of user sequences. Then we combine the cloze task mask and the dropout mask to generate high-quality positive samples and perform multi-pair contrastive learning, which demonstrates better performance and adaptability compared with the normal one-pair contrastive learning. Moreover, we introduce a novel dynamic loss reweighting strategy to balance between the cloze task loss and the contrastive loss. Experiment results on three public benchmark datasets show that our model outperforms state-of-the-art models for sequential recommendation.