Intelligent Anomaly Detection in Social Media
Yanwei Xu, Zhiyong Feng, Schahram Dustdar · 2025
Social media anomaly detection is a crucial component for enhancing the overall integrity and functionality of social media platforms. Its role is multifaceted, including key aspects such as security and fraud detection, where it plays a pivotal role in identifying and mitigating potential risks like fake accounts, malicious attacks, and cyberbullying. With the increasing popularity of social media, a plethora of user-generated content, including images, videos, and comments, is showcased on online social platforms. Consequently, anomaly detection on social media encounters various challenges such as data diversity, real-time requirements, and privacy preservation. To address these challenges, researchers have developed traditional and machine learning-based intelligent anomaly detection methods tailored to various types of anomalous activities. This work primarily focuses on anomalies and abnormal behaviours on online social platforms, providing a comprehensive review of recent developments in intelligent detection methods. We present a broad overview covering the background, common terminology, comparisons between traditional and intelligent approaches, application scenarios, and potential research directions. In this endeavour, we aim to contribute valuable theoretical and technical support for the future innovation and development of anomaly detection in social media.