Seed selection for information cascade in multilayer networks

TL;DR

This paper investigates how seed selection strategies influence information spread in multilayer social networks, revealing that traditional single-layer methods underperform compared to Degree Centrality in multilayer contexts.

Contribution

It evaluates existing seed selection strategies on multilayer networks, highlighting the superior performance of Degree Centrality over methods like K-Shell and VoteRank.

Findings

Degree Centrality outperforms other strategies in multilayer networks.

Traditional single-layer seed selection methods are less effective in multilayer settings.

Analysis conducted on eighteen real-world multilayer social networks.

Abstract

Information spreading is an interesting field in the domain of online social media. In this work, we are investigating how well different seed selection strategies affect the spreading processes simulated using independent cascade model on eighteen multilayer social networks. Fifteen networks are built based on the user interaction data extracted from Facebook public pages and tree of them are multilayer networks downloaded from public repository (two of them being Twitter networks). The results indicate that various state of the art seed selection strategies for single-layer networks like K-Shell or VoteRank do not perform so well on multilayer networks and are outperformed by Degree Centrality.