Deep Identification of Propagation Trees

2025-03-01

AI Summary

Reconstructing the propagation tree—i.e., inferring “who infected whom”—during graph diffusion processes is essential for understanding underlying transmission mechanisms; however, existing methods focus solely on source localization and cannot recover the full propagation structure. To address this, we propose DIPT, an end-to-end probabilistic framework that jointly models node-level local influence and propagation-tree topology, enabling coupled inversion of both diffusion mechanisms and graph structure. DIPT employs differentiable probabilistic modeling and alternating optimization, requiring no predefined diffusion model or prior structural assumptions. Evaluated on five real-world datasets, DIPT significantly outperforms state-of-the-art baselines, achieving an average 12.6% improvement in propagation-tree reconstruction F1-score. This work breaks the long-standing limitation of decoupled source localization and path inference, establishing a unified approach to mechanistic diffusion analysis on graphs.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Graph-based Machine Learning
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• Web Mining and Content Analysis: Content-based information diffusion

Abstract

Understanding propagation structures in graph diffusion processes, such as epidemic spread or misinformation diffusion, is a fundamental yet challenging problem. While existing methods primarily focus on source localization, they cannot reconstruct the underlying propagation trees i.e.,"who infected whom", which are substantial for tracking the propagation pathways and investigate diffusion mechanisms. In this work, we propose Deep Identification of Propagation Trees (DIPT), a probabilistic framework that infers propagation trees from observed diffused states. DIPT models local influence strengths between nodes and leverages an alternating optimization strategy to jointly learn the diffusion mechanism and reconstruct the propagation structure. Extensive experiments on five real-world datasets demonstrate the effectiveness of DIPT in accurately reconstructing propagation trees.

Problem

• Reconstructing propagation trees in graph diffusion processes.
• Inferring 'who infected whom' in epidemic or misinformation spread.
• Learning diffusion mechanisms and local influence strengths between nodes.

Innovation

• Probabilistic framework infers propagation trees
• Models local influence strengths between nodes
• Alternating optimization learns diffusion mechanisms