Temporal Linear Item-Item Model for Sequential Recommendation (WSDM'25)

This is the official code for TALE in the paper "Temporal Linear Item-Item Model for Sequential Recommendation", The 18th ACM International Conference on Web Search and Data Mining.

Overview

TALE is a novel linear sequential recommendation approach that effectively captures temporal dynamics in user-item interactions. Unlike complex neural models, TALE achieves superior performance while maintaining high computational efficiency through its linear item-item formulation. Our model addresses two key limitations in sequential recommendation:

• Efficiency: TALE's linear architecture provides dramatic speedups (up to 57× faster) compared to transformer-based models
• Temporal modeling: Unlike existing methods that only consider sequential ordering, TALE leverages actual timestamps to model time-dependent preference shifts

The key innovations of our approach include:

• Single-target augmentation: Focuses on a single target item to learn precise temporal correlations
• Time interval-aware weighting: Utilizes actual timestamps to model item relationships based on time intervals
• Trend-aware normalization: Reflects dynamic changes in item popularity over time

Requirements

We will update the complete implementation environments soon.

How to run

python run_recbole.py --dataset=beauty --model=TALE --reg_weight=200 --tau_train=0.5 --tau_inf=2

Citation

Please cite our paper if you find our work helpful.

@inproceedings{park2025tale,
  title={Temporal Linear Item-Item Model for Sequential Recommendation},
  author={Seongmin Park and
          Mincheol Yoon and
          Minjin Choi and
          Jongwuk Lee},
  booktitle={The 18th ACM International Conference on Web Search and Data Mining (WSDM)},
  year={2025}
}