Fig 1.
Overall architecture of TFP-Net.
N: Number of customers, T: Number of time steps, d: Embedding dimension, df: Dimension of the fused feature vector (), C: Number of customer segments. The framework processes raw data to generate temporal embeddings of shape (N, T, d), which are fused into static features to form vectors of shape (N, df). These features are fed into DeepFM and further processed by ProtoNet, which learns C prototypes of shape (C, df) to produce the final prediction output of shape (N, C).
Fig 2.
Graph representation of customer interactions after processing with the TGAT model, where each node represents a customer and edges indicate interactions between customers over time.
Table 1.
Comparison results of TFP-Net and baseline models on taobao user behavior and Amazon Product datasets (mean ± std over 5 independent runs and 5-fold cross-validation).
Table 2.
Comparison of model efficiency on taobao user behavior and amazon product datasets.
Table 3.
Comparison of model performance in cold-start scenarios (1-shot, 5-shot, 10-shot). Results are reported as mean ± std over 5 independent runs with different random seeds.
Fig 3.
Attention weight distributions.
(a) Taobao: Time-Sensitive Behaviors — Attention weights for different customer behaviors around a promotional event. (b) Amazon: Product Features — Attention weights for product features across different categories.
Fig 4.
t-SNE Visualization of Customer and Product Prototypes.
(a) Taobao: Prototypes by Behavior Patterns — t-SNE visualization of customer prototypes based on behavior patterns. (b) Amazon: Prototypes by Product Preferences — t-SNE visualization of user prototypes based on product preferences.
Table 4.
TFP-Net core hyperparameter sensitivity analysis results. Results are reported as mean ± std over 5 independent runs with different random seeds.
Table 5.
Ablation study results of TFP-Net components. Results are reported as mean ± std over 5 independent runs.