Significant Ties Graph Neural Networks for Continuous-Time Temporal Networks Modeling

TL;DR

This paper introduces Significant Ties Graph Neural Networks (STGNN), a framework that effectively models continuous-time temporal networks by capturing significant interactions and adaptively weighting node relationships, outperforming snapshot-based methods.

Contribution

The paper presents a novel STGNN framework that models significant ties in continuous-time temporal networks, improving over traditional snapshot-based approaches.

Findings

STGNN outperforms existing methods on four real networks.

The aggregation mechanism effectively captures significant historical interactions.

Adaptive significance weighting enhances modeling of dynamic network evolution.

Abstract

Temporal networks are suitable for modeling complex evolving systems. It has a wide range of applications, such as social network analysis, recommender systems, and epidemiology. Recently, modeling such dynamic systems has drawn great attention in many domains. However, most existing approaches resort to taking discrete snapshots of the temporal networks and modeling all events with equal importance. This paper proposes Significant Ties Graph Neural Networks (STGNN), a novel framework that captures and describes significant ties. To better model the diversity of interactions, STGNN introduces a novel aggregation mechanism to organize the most significant historical neighbors' information and adaptively obtain the significance of node pairs. Experimental results on four real networks demonstrate the effectiveness of the proposed framework.