Learnable Sequence Augmenter for Triplet Contrastive Learning in Sequential Recommendation

TL;DR

This work tackles data sparsity and noise in sequential recommendation by introducing LACLRec, which replaces random augmentations with a learnable SSL sequence augmenter that deletes noisy items and inserts related ones. It couples this augmenter with a two-pronged contrastive objective, combining a coarse $L_{cl}$ with a finer $L_{tri}$ to produce richer supervision and a more fine-grained alignment between raw sequences and augmented views. Extensive experiments on Beauty, Yelp, and Sports show that LACLRec surpasses baselines including CL4SRec, with notable gains attributed to the high-quality SSL-augmented sequences and the triplet training signal. The method also demonstrates robustness to noisy interactions and provides insights into hyperparameter sensitivities, highlighting practical considerations for deploying learnable augmentation in SR systems.

Abstract

Most existing contrastive learning-based sequential recommendation (SR) methods rely on random operations (e.g., crop, reorder, and substitute) to generate augmented sequences. These methods often struggle to create positive sample pairs that closely resemble the representations of the raw sequences, potentially disrupting item correlations by deleting key items or introducing noisy iterac, which misguides the contrastive learning process. To address this limitation, we propose Learnable sequence Augmentor for triplet Contrastive Learning in sequential Recommendation (LACLRec). Specifically, the self-supervised learning-based augmenter can automatically delete noisy items from sequences and insert new items that better capture item transition patterns, generating a higher-quality augmented sequence. Subsequently, we randomly generate another augmented sequence and design a ranking-based triplet contrastive loss to differentiate the similarities between the raw sequence, the augmented sequence from augmenter, and the randomly augmented sequence, providing more fine-grained contrastive signals. Extensive experiments on three real-world datasets demonstrate that both the sequence augmenter and the triplet contrast contribute to improving recommendation accuracy. LACLRec significantly outperforms the baseline model CL4SRec, and demonstrates superior performance compared to several state-of-the-art sequential recommendation algorithms.

Figures (2)

• Figure 1: Illustration of triplet contrastive learning.
• Figure 2: Overview of LACLRec. When training augmenter, we require it to restore the random modified sequences. The augmented sequence $S_u^{aug1}$ is generated by the augmenter, and $S_u^{aug2}$ is generated by random methods. Two contrastive learning tasks are used to provide supervised signals for item correlations learning. Finally, the recommender predicts the next item.