Local Community-Based Anomaly Detection in Graph Streams

The problem of anomaly detection on static networks has been broadly studied in various research domains. Anomaly detection concerns the identification of objects or connections between objects that differ substantially from the rest. However, a plethora of real-world networks are highly dynamic, in the sense that entities (nodes) as well as relations between them (edges) constantly change. The time dimension critically affects the very definition of an anomaly in these cases, and thus care must be taken to properly incorporate it. Many solutions have been proposed in dynamic/temporal networks under various assumptions concerning the modeling of time as well as the detected anomalies. The problem becomes quite harder when the notion of time is introduced since various unseen problems arise when compared to the static case. Our study aims to provide a comprehensive examination of recent developments in community-based anomaly detection and to discuss preliminary notions related to temporal networks. Furthermore, we introduce an established algorithm for dynamic local community detection, aimed at identifying anomalies based on community evolution over time. We also provide preliminary experimental findings using synthetic datasets to support our study.