Basser Seminar Series
Finding anomalies in graph/relational data
Speaker: Professor David B. Skillicorn
School of Computing, Queen's University, Kingston, Canada
Time: Wednesday 4 June 2008, 4-5pm
Location: The University of Sydney, School of IT Building, Lecture Theatre (Room 123), Level 1
Graph or relational data capture emergent properties, and so are particularly powerful and resistant to manipulation in adversarial knowledge discovery settings, such as policing and counterterrorism. However, such data is difficult to work with: existing technologies either use rendering techniques, which tend to emphasise regularities rather than anomalies, or exploration from a single node, which requires prior information about which nodes are interesting.
Embedding graphs in Euclidean space using spectral techniques is one way to get the benefits both of relational structure and useful geometry. However, usually only a few eigenvectors or dimensions of such an embedding are considered. I will show that anomalous regions can be discovered without prior information by considering `middle' eigenvectors. I will illustrate using relational data about al Qaeda.
David Skillicorn (email@example.com) is a Professor in the School of Computing at Queen's University in Canada. His research is in smart information management, both the problems of extracting and sharing useful knowledge from data, and the problems of accessing and computing with large datasets that are geographically distributed. He has published extensively in high-performance computing and data mining.
At present his focus is on understanding complex datasets in applications such as biomedicine, geochemistry, network intrusion, fraud detection, and counterterrorism. He has an undergraduate degree from the University of Sydney and a Ph.D. from the University of Manitoba.