Abstract:Sequential recommendation typically utilizes deep neural networks to mine rich information in interaction sequences. However, existing methods often face the issue of insufficient interaction data. To alleviate the sparsity issue, self-supervised learning is introduced into sequential recommendation. Despite its effectiveness, we argue that current self-supervised learning-based (i.e., SSL-based) sequential recommendation models have the following limitations: (1) using only a single self-supervised learning method, either contrastive self-supervised learning or generative self-supervised learning. (2) employing a simple data augmentation strategy in either the graph structure domain or the node feature domain. We believe that they have not fully utilized the capabilities of both self-supervised methods and have not sufficiently explored the advantages of combining graph augmentation schemes. As a result, they often fail to learn better item representations. In light of this, we propose a novel multi-task sequential recommendation framework named Adaptive Self-supervised Learning for sequential Recommendation (ASLRec). Specifically, our framework combines contrastive and generative self-supervised learning methods adaptively, simultaneously applying different perturbations at both the graph topology and node feature levels. This approach constructs diverse augmented graph views and employs multiple loss functions (including contrastive loss, generative loss, mask loss, and prediction loss) for joint training. By encompassing the capabilities of various methods, our model learns item representations across different augmented graph views to achieve better performance and effectively mitigate interaction noise and sparsity. In addition, we add a small proportion of random uniform noise to item representations, making the item representations more uniform and mitigating the inherent popularity bias in interaction records. We conduct extensive experiments on three publicly available benchmark datasets to evaluate our model. The results demonstrate that our approach achieves state-of-the-art performance compared to 14 other competitive methods: the hit rate (HR) improved by over 14.39%, and the normalized discounted cumulative gain (NDCG) increased by over 18.67%.

Graph Neural Network-Guided Contrastive Learning for Sequential Recommendation

Graph Contrastive Learning with Adaptive Augmentation for Recommendation.

Adaptive Graph Contrastive Learning for Recommendation

Enhancing Sequential Recommendation with Graph Contrastive Learning

Explanation Guided Contrastive Learning for Sequential Recommendation

GNNCL: A Graph Neural Network Recommendation Model Based on Contrastive Learning

Enhancing Graph Contrastive Learning with Reliable and Informative Augmentation for Recommendation

Enhancing sequential recommendation with contrastive Generative Adversarial Network

Self-Attention Based Sequential Recommendation with Graph Convolutional Networks

Graph convolutional network and self-attentive for sequential recommendation

Adaptive Fusion of Multi-View for Graph Contrastive Recommendation

Contrastive Learning with Reverse Extended Enhancement for Sequential Recommendation

LightGCL: Simple Yet Effective Graph Contrastive Learning for Recommendation

Graph Convolutional Networks with Collaborative Feature Fusion for Sequential Recommendation

SACL: Siamese Adaptive Contrastive Learning for Recommendation

Dual Contrastive Network for Sequential Recommendation

Adaptive self-supervised learning for sequential recommendation

Simplified Graph Contrastive Learning for Recommendation with Direct Optimization of Alignment and Uniformity

Quaternion-Based Graph Contrastive Learning for Recommendation

Globally Informed Graph Contrastive Learning for Recommendation

Contrastive Graph Learning for Social Recommendation