Adaptive Cross-Platform Threat Correlation Layer (ACTCL):A Cybersecurity Approach to Detecting Distributed Digital Violence on Social Media Platforms

Abstract

Large-scale digital assaults and online harassment increasingly span multiple social media platforms, yet cross-platform threat correlation remains limited by privacy regulations, incompatible data formats, and the absence of a shared behavioural analysis layer. Current privacy-conscious machine-learning solutions focus on platform-specific anomaly detection rather than on correlating behaviour across decoupled environments.

This paper introduces the Adaptive Cross-Platform Threat Correlation Layer (ACTCL), a lightweight cybersecurity framework that reframes digital violence as a privacy-aware behavioural correlation problem. ACTCL employs Anonymous Feature Hashing (AFH) to convert three transient behavioural features—temporal rhythm, activity concentration, and session duration—into irreversible eight-dimensional fingerprints that can be compared across systems without exposing raw data. Using a controlled synthetic dataset of 150 users, including 10 coordinated attackers, cosine similarity over hashed fingerprints reveals a degenerate similarity cluster for coordinated users (mean = 1.0000, SD = 0.0000) and a more dispersed distribution for normal users (mean = 0.9877, SD = 0.0224).

These results show that organised malicious behaviour retains stable behavioural signatures even under strong anonymisation, enabling privacy-compliant cross-platform threat correlation. ACTCL offers a practical foundation for future multi-platform security cooperation, with applications in fraud detection, coordinated misinformation tracking, botnet analysis, and other distributed digital threats.