Dynamic Network Discovery via Infection Tracing

Ben Bals,Michelle Döring,Nicolas Klodt,George Skretas

2024-12-15

Abstract:Researchers, policy makers, and engineers need to make sense of data from spreading processes as diverse as rumor spreading in social networks, viral infections, and water contamination. Classical questions include predicting infection behavior in a given network or deducing the network structure from infection data. Most of the research on network infections studies static graphs, that is, the connections in the network are assumed to not change. More recently, temporal graphs, in which connections change over time, have been used to more accurately represent real-world infections, which rarely occur in unchanging networks. We propose a model for temporal graph discovery that is consistent with previous work on static graphs and embraces the greater expressiveness of temporal graphs. For this model, we give algorithms and lower bounds which are often tight. We analyze different variations of the problem, which make our results widely applicable and it also clarifies which aspects of temporal infections make graph discovery easier or harder. We round off our analysis with an experimental evaluation of our algorithm on real-world interaction data from the Stanford Network Analysis Project and on temporal Erdős-Renyi graphs. On Erdős-Renyi graphs, we uncover a threshold behavior, which can be explained by a novel connectivity parameter that we introduce during our theoretical analysis.

Data Structures and Algorithms

What problem does this paper attempt to address?

### What problem does this paper attempt to solve?

This paper aims to solve the problem of **Temporal Graph Discovery (TGD)**, especially how to infer network structures through infection tracking. Specifically, the author focuses on how to discover the underlying network structures and their time labels from infection data in **temporal graphs**.

#### Background and Motivation

1. **Limitations of Static Graphs**: Most previous studies have focused on static graphs, that is, assuming that the connections in the network are fixed. However, in the real world, many spreading processes (such as disease spread, rumor diffusion, etc.) occur in constantly changing networks. Therefore, the static graph model cannot accurately capture these dynamic characteristics.

2. **Advantages of Temporal Graphs**: Temporal graphs allow the existence of edges to change over time, which is more in line with the actual situation. For example, the social connection between two people may only exist during a specific period of time, rather than all the time.

3. **Research Gap**: Although temporal graphs have been widely used in other fields, there has been no research specifically for temporal graphs in network discovery tasks. This paper fills this gap and proposes a network discovery model and algorithm suitable for temporal graphs.

#### Main Contributions

1. **Defining the TGD Problem**: The author models temporal graph discovery as a turn - based interactive game, in which one party (Discoverer, D) tries to infer the network structure through infection data, and the other party (Adversary, A) tries to delay D's progress by controlling the infection behavior.

2. **Proposing Algorithms and Theoretical Analysis**:
- Proposes the `DiscoveryFollow` algorithm, which can solve the graph discovery problem within \(6|E(G)| + |EC_δ(G)| \lceil T_{\max}/δ \rceil\) rounds, where \( |EC_δ(G) |\) is the number of δ - edge - connected components.
- Provides a theoretical proof, showing that the time complexity of this algorithm is asymptotically optimal in terms of the number of edges.
- Proves the lower bounds of certain graph categories, showing that even if D is allowed to start multiple infection chains per round, the performance cannot be significantly improved.

3. **Experimental Verification**: Through experiments on real - world datasets (such as social network data in the Stanford Network Analysis Project) and synthetic data (such as temporal Erdős - Rényi graphs), the effectiveness of the proposed algorithm is verified, and the factors affecting the algorithm performance are analyzed.

#### Key Concepts

- **Temporal Graph**: Consists of a set of nodes \( V \), a set of edges \( E \), and an edge time - label function \( λ:E \to 2^{[T_{\max}]} \), indicating that each edge exists only at specific time steps.

- **δ - edge - connected components (δ - ecc)**: A connectivity parameter in temporal graphs, reflecting the time constraints between edges, so that the infection chains must follow these time limits.

- **SIR Model**: A model used to simulate infection behavior, where nodes can be in one of three states: susceptible, infected, or resistant.

### Summary

The main goal of this paper is to develop a new framework and algorithm to deal with the network discovery problem in temporal graphs. By introducing the temporal graph model, the author not only extends the existing static graph research but also provides a powerful tool for understanding and dealing with dynamic spreading processes in the real world.