Bandpass Threshold Models for Activation and Influence Propagation in Social Networks
We consider the models of activation/influence propagation in social networks based on the concept of “bandpass” thresholds: a node will “activate” if at least a certain minimum fraction of and no more than a certain maximum fraction of its neighbors are a...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11075656/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | We consider the models of activation/influence propagation in social networks based on the concept of “bandpass” thresholds: a node will “activate” if at least a certain minimum fraction of and no more than a certain maximum fraction of its neighbors are active. These interval-based activation processes are conceptually similar to a “bandpass filter” in signal processing, implying that activation occurs only for a signal in a certain range between a minimum and a maximum frequency values. The existence of such phenomena in social networks was originally proposed in the seminal work on threshold models by Granovetter (1978); however, these models have not yet been extensively studied from mathematical modeling and computational perspectives. In this study, we investigate two baseline versions of these models by conducting computational experiments on multiple publicly available social networks (with various sizes, edge densities, and clustering coefficients), as well as on very large (city-wide and country-wide) online social media networks constructed based on anonymized data collected from a well-known social media portal VK.com in December 2016. We also compare the results to the ones for classical linear threshold (“single-threshold”) models with the same lower threshold values. We observe that the processes of bandpass threshold activation/influence propagation can exhibit drastically different behaviors depending on initial choices of seed nodes, underlying network characteristics, and lower/upper threshold values. |
|---|---|
| ISSN: | 2169-3536 |