ACM WSDM 2013 Tutorial

Anomaly, Event, and Fraud Detection in Large Graph Datasets

Leman Akoglu

Stony Brook University
Department of Computer Science

Christos Faloutsos

Carnegie Mellon University
School of Computer Science



Slides (see Proposal Outline)


Detecting anomalies and events in data is a vital task, with numerous applications in security, finance, health care, law enforcement, and many others. While many techniques have been developed in past years for spotting outliers and anomalies in unstructured collections of multi-dimensional points, with graph data becoming ubiquitous, techniques for structured graph data have been of focus recently.

The goal of this tutorial is to provide a general, comprehensive overview of the state-of-the-art methods for anomaly, event, and fraud detection in data represented as graphs. As a key contribution, we provide a thorough exploration of both data mining and machine learning algorithms for these detection tasks. We give a general framework for the algorithms, categorized under various settings: unsupervised vs. (semi-)supervised, for static vs. dynamic data. We focus on the scalability and effectiveness aspects of the methods, and highlight results on crucial real-world applications, including accounting fraud and opinion spam detection.

List of references

The following publications are referenced in the tutorial (categorized by each major topic).


Outlier and Anomaly detection

Outlier detection in clouds of multi-dimensional points:

Anomaly detection in graph data:


Event/Outbreak, and Fraud detection

Event/Outbreak detection:

  Relational Learning with networks

Links to talks/tutorials by tutors

Contact information

Christos Faloutsos
Carnegie Mellon University,
School of Computer Science
GHC 8019 Pittsburgh, PA 15213
Leman Akoglu
Stony Brook University,
Department of Computer Science
1425 CS Bldg. Stony Brook, NY 11794