Core Concepts
Follow back accounts manipulate social media platforms by inflating follower counts, potentially causing harm through misinformation. This study provides insights into their characteristics and behaviors.
Abstract
Abstract:
First large-scale analysis of follow back accounts.
Identified 12 communities from 12 countries.
Proposed a classifier for detecting follow back accounts.
Introduction:
Popular social media users employ growth strategies.
Follow back accounts manipulate follower counts.
Summary of Findings:
Discovered 2759 follow back accounts and 12 communities.
Proposed a classifier with moderate success in classification.
Related Work:
Extensive research on manipulation strategies in social media.
Definition:
Defined follow back behavior as reciprocal followings to inflate follower counts.
Data Collection:
Used honeypot approach to collect ground truth data on follow back accounts.
Communities of Follow Back Accounts:
Identified characteristics and behaviors of different communities based on country or interest.
Characterization:
Analyzed differences between follow back and other accounts in terms of activity, engagement, reciprocity, etc.
Platform Abuse:
Examined coordination, automation, trains conduct/ride behavior among communities.
Suspensions:
Low suspension rate for follow back accounts despite T.O.S. violations.
Stats
"We discovered and describe 12 communities of follow"
"We found that a tabular data classifier using features created from profile metadata"
"Our initial ground truth dataset consists of"
Quotes
"We propose a classifier for such accounts and report that models employing profile metadata and the ego network demonstrate promising results."
"Despite their potential harm, such accounts are understudied. We fill this gap and present the first large-scale study of follow back accounts."