SS4Rec: Continuous-Time Sequential Recommendation with State Space Models

TL;DR

This work tackles the challenge of irregular time intervals in sequential recommendations by formulating the problem as a continuous-time system and proposing SS4Rec, a hybrid model that unites time-aware and relation-aware state space models with variable-step discretization. By encoding temporal context through a time-aware SSM and selective relational dynamics through a Mamba-style SSM, SS4Rec achieves time-specific, accurate recommendations while maintaining efficient inference. Empirical results on five real-world datasets demonstrate superior performance over strong baselines, along with thorough ablations, time-dependent predictions, and partial-observation robustness. The approach offers practical impact for real-world systems requiring fine-grained, time-aware user modeling and scalable sequential reasoning.

Abstract

Sequential recommendation is a key area in the field of recommendation systems aiming to model user interest based on historical interaction sequences with irregular intervals. While previous recurrent neural network-based and attention-based approaches have achieved significant results, they have limitations in capturing system continuity due to the discrete characteristics. In the context of continuous-time modeling, state space model (SSM) offers a potential solution, as it can effectively capture the dynamic evolution of user interest over time. However, existing SSM-based approaches ignore the impact of irregular time intervals within historical user interactions, making it difficult to model complexed user-item transitions in sequences. To address this issue, we propose a hybrid SSM-based model called SS4Rec for continuous-time sequential recommendation. SS4Rec integrates a time-aware SSM to handle irregular time intervals and a relation-aware SSM to model contextual dependencies, enabling it to infer user interest from both temporal and sequential perspectives. In the training process, the time-aware SSM and the relation-aware SSM are discretized by variable stepsizes according to user interaction time intervals and input data, respectively. This helps capture the continuous dependency from irregular time intervals and provides time-specific personalized recommendations. Experimental studies on five benchmark datasets demonstrate the superiority and effectiveness of SS4Rec.

Figures (5)

• Figure 1: Overview of the SS4Rec framework.
• Figure 2: Experiments on Toy dataset. Time-interval embedding is added to item embedding for baselines.
• Figure 3: Operation of dropping some items in sequence.
• Figure 4: Comparison of practical time/space cost.
• Figure 5: Experiments of different lengths on KuaiRec(Large) dataset.