Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

#TeamFollowBack: Detection & Analysis of Follow Back Accounts on Social Media

Tuğrulcan Elmas, Mathis Randl, Youssef Attia

TL;DR

Follow back accounts inflate follower counts and distort perceived influence on social platforms. The authors introduce the first large-scale analysis of follow back behavior on X using a honeypot-based ground-truth approach, uncovering 12 communities across 12 countries with distinct political and commercial orientations. They show that follow back accounts tend to be younger, more engaged, and have inflated follower/following counts, and they develop classifiers based on profile metadata and ego networks with moderate recall, though cross-community generalization remains challenging. The work highlights implications for platform integrity and civic discourse, and provides methodological guidance for detecting coordinated follow back behavior while acknowledging ethical and data-collection limitations.

Abstract

Follow back accounts inflate their follower counts by engaging in reciprocal followings. Such accounts manipulate the public and the algorithms by appearing more popular than they really are. Despite their potential harm, no studies have analyzed such accounts at scale. In this study, we present the first large-scale analysis of follow back accounts. We formally define follow back accounts and employ a honeypot approach to collect a dataset of such accounts on X (formerly Twitter). We discover and describe 12 communities of follow back accounts from 12 different countries, some of which exhibit clear political agenda. We analyze the characteristics of follow back accounts and report that they are newer, more engaging, and have more followings and followers. Finally, we propose a classifier for such accounts and report that models employing profile metadata and the ego network demonstrate promising results, although achieving high recall is challenging. Our study enhances understanding of the follow back accounts and discovering such accounts in the wild.

#TeamFollowBack: Detection & Analysis of Follow Back Accounts on Social Media

TL;DR

Follow back accounts inflate follower counts and distort perceived influence on social platforms. The authors introduce the first large-scale analysis of follow back behavior on X using a honeypot-based ground-truth approach, uncovering 12 communities across 12 countries with distinct political and commercial orientations. They show that follow back accounts tend to be younger, more engaged, and have inflated follower/following counts, and they develop classifiers based on profile metadata and ego networks with moderate recall, though cross-community generalization remains challenging. The work highlights implications for platform integrity and civic discourse, and provides methodological guidance for detecting coordinated follow back behavior while acknowledging ethical and data-collection limitations.

Abstract

Follow back accounts inflate their follower counts by engaging in reciprocal followings. Such accounts manipulate the public and the algorithms by appearing more popular than they really are. Despite their potential harm, no studies have analyzed such accounts at scale. In this study, we present the first large-scale analysis of follow back accounts. We formally define follow back accounts and employ a honeypot approach to collect a dataset of such accounts on X (formerly Twitter). We discover and describe 12 communities of follow back accounts from 12 different countries, some of which exhibit clear political agenda. We analyze the characteristics of follow back accounts and report that they are newer, more engaging, and have more followings and followers. Finally, we propose a classifier for such accounts and report that models employing profile metadata and the ego network demonstrate promising results, although achieving high recall is challenging. Our study enhances understanding of the follow back accounts and discovering such accounts in the wild.
Paper Structure (18 sections, 3 figures, 3 tables)

This paper contains 18 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Summary of Findings:
  3. Related Work
  4. Definition
  5. Data Collection
  6. Ground Truth
  7. Sampling Methodology
  8. Communities of Follow Back Accounts
  9. Semantic Characteristics of Communities
  10. Characterization
  11. Platform Abuse
  12. Classification
  13. Models Based on Single Account Data
  14. Context-Aware (Network) Models
  15. Experimental Details & Results
  16. ...and 3 more sections

Figures (3)

  • Figure 1: Communities, ordered by their size.
  • Figure 2: The characteristics vs. follow back ratios. The trend lines are computed using linear regression.
  • Figure 3: Level of coordination among communities: The communities depicted in the left plot exhibit a high number of weakly coordinated users but a low number of highly coordinated users. In contrast, the communities shown in the right plot have a relatively high number of highly correlated users but a low overall count of coordinated users.