Rumour Spreading Depends on the Latent Geometry and Degree Distribution in Social Network Models

Marc Kaufmann,Kostas Lakis,Johannes Lengler,Raghu Raman Ravi,Ulysse Schaller,Konstantin Sturm

2024-08-07

Abstract:We study push-pull rumour spreading in small-world models for social networks where the degrees follow a power-law. In a non-geometric setting Fountoulakis, Panagiotou and Sauerwald have shown that rumours always spread fast (SODA 2012). On the other hand, Janssen and Mehrabian have found that rumours spread slowly in a spatial preferential attachment model (SIDMA 2017). We study the question systematically for the model of geometric inhomogeneous random graphs (GIRGs), which has been found to be a good theoretical and empirical fit for social networks. Our result is two-fold: with classical Euclidean geometry both slow and fast rumour spreading may occur, depending on the exponent of the power law and the prevalence of weak ties in the networks, and we fully characterise the phase boundaries between those two regimes. Depending on the parameters, fast spreading may either mean polylogarithmic time or even doubly logarithmic time. Secondly, we show that rumour spreading is always fast in a non-metric geometry. The considered non-metric geometry allows to model social connections where resemblance of vertices in a single attribute, such as familial kinship, already strongly indicates the presence of an edge. Classical Euclidean Geometry fails to capture such ties. For some regimes in the Euclidean setting, the efficient pathways for spreading rumours differ from previously identified paths. A vertex of degree $d$ can transmit the rumour efficiently to a vertex of larger degree by a chain of length $3$, where one of the two intermediaries has constant degree, and the other has degree $d^{c}$ for some constant $c<1$.

Probability,Social and Information Networks,Combinatorics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **How does the spread speed of rumors in the social network model depend on the underlying geometric structure and degree distribution?**

Specifically, the authors studied the behavior of the push - pull rumor - spreading mechanism in the small - world social network model with a power - law degree distribution. They paid special attention to the Geometric Inhomogeneous Random Graphs (GIRGs) model, which is considered to be a good theoretical and empirical fit for social networks. By systematically studying the GIRGs model, the authors aimed to answer the following questions:

1. **Is the speed of rumor spreading affected by the underlying geometric structure?**
2. **Under different parameter settings (such as the power - law exponent τ and the prevalence of weak connections), is the rumor spreading fast or slow?**

### Main Findings

- **GIRGs under Euclidean Geometry**:
- When the power - law exponent τ and the existence of weak connections satisfy certain conditions, rumors can spread quickly or super - quickly.
- For example, if α < τ−1 / (τ−2), rumors can spread quickly in polylogarithmic time; if τ < √2 + 1, they can spread super - quickly in double - logarithmic time.
- If τ > ϕ + 1 and α > τ−1 / (τ−2), the rumor spreading will be slow and it will take polynomial time to spread to a large number of nodes.

- **MCD - GIRGs under Non - Euclidean Geometry**:
- In the Minimum Component Distance (MCD) geometry, rumors always spread quickly.
- When the typical distance is ultrasmall, the rumor spreading is even super - quickly.

### Formula Summary

- **Connection Probability Formula**:
\[
p_{uv} = \Theta\left(\min\left\{\left(\frac{W_u W_v}{n V(|x_u - x_v|)}\right)^\alpha, 1\right\}\right)
\]
where \(W_u\) and \(W_v\) are the weights of nodes \(u\) and \(v\) respectively, \(V(|x_u - x_v|)\) is the volume function induced by the distance function, and α is the parameter that controls the prevalence of weak connections.

- **Spread Time Formula**:
- The time complexity of fast spreading is \((\log n)^{\Delta(\gamma)+o(1)}\), where \(\Delta(\gamma) = 1 / \log_2(2 / \gamma)\).
- The time complexity of super - fast spreading is \((4 + o(1))\cdot\frac{\log\log n}{|\log(\alpha(\tau - 2))|}\) or \((6 + o(1))\cdot\frac{\log\log n}{|\log(\tau(\tau - 2))|}\).

### Conclusion

This research reveals that the rumor - spreading speed in social networks depends not only on the network's topological structure, but also is significantly affected by the underlying geometric structure and degree distribution. In particular, different power - law exponents and the prevalence of weak connections can lead to huge changes in rumor spreading from slow to fast or even super - fast. These findings are of great significance for understanding the information - spreading behavior in networks and provide a theoretical basis for designing more effective information - spreading strategies.