Beyond Isolated Clients: Integrating Graph-Based Embeddings into Event Sequence Models

arXiv cs.LG / 4/13/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper argues that existing self-supervised learning for event sequence modeling captures temporal order well but largely ignores the broader user–item interaction graph structure.
It proposes three model-agnostic strategies to inject graph-based information into contrastive SSL: enriching event embeddings, aligning client representations with graph embeddings, and introducing a structural pretext task.
Experiments across four financial and e-commerce datasets show consistent accuracy gains, with improvements reported up to 2.3% AUC.
The authors find that interaction graph density strongly influences which integration strategy performs best, guiding practical model design choices.

Abstract

Large-scale digital platforms generate billions of timestamped user-item interactions (events) that are crucial for predicting user attributes in, e.g., fraud prevention and recommendations. While self-supervised learning (SSL) effectively models the temporal order of events, it typically overlooks the global structure of the user-item interaction graph. To bridge this gap, we propose three model-agnostic strategies for integrating this structural information into contrastive SSL: enriching event embeddings, aligning client representations with graph embeddings, and adding a structural pretext task. Experiments on four financial and e-commerce datasets demonstrate that our approach consistently improves the accuracy (up to a 2.3% AUC) and reveals that graph density is a key factor in selecting the optimal integration strategy.