MathSys 2021-22 MSc group project at University of Warwick.
Our external partners at Kirontech offer a Health Insurance Platform that helps insurance payers deal with anomalies in their insurance claims. Healthcare data naturally involves a number of relevant interactions between different entities. Kirontech has yet to explore graph-based anomaly detection (GBAD) techniques.
The goal of this project is to adopt different GBAD methods on Kirontech's real-life dataset and provide evidence that graph-based data is useful to examine different anomalies.
- Fraud detection: A systematic literature review of graph-based anomaly detection approaches - Tahereh Pourhabibi et al (2020)
- CADA: A Community-Aware Approach for Identifying Node Anomalies in Complex Networks - TJ Helling, JC Scholtes, FW Takes (2018)
- Community Detection - Spectral Bipartitioning & Modularity Maximization: Finding community structure in networks using the eigenvectors of matrices - M. E. J. Newman (2006)
- Modularity Maximization - Leiden Algorithm:
- From louvain to leiden: guaranteeing well-connected communities - TV A, W L, van Eck N J (2019)
- Narrow scope for resolution-limit-free community detection - VA Traag, PV Dooren, Y Nesterov (2011)
- Modularity Maximization - Robustness: Dynamic communities in multichannel data: An application to the foreign exchange market during the 2007-2008 credit crisis - DJ Fenn, et al (2009)
- Modularity Maximization - Leiden Algorithm:
- PSS Algorithm: Detecting Healthcare Fraud through Patient Sharing Schemes - Aryya Gangopadhyay, Song Chen, and Yelena Yesha (2012)
- OddBall: OddBall: Spotting Anomalies in Weighted Graphs - Leman Akoglu, Mary McGlohon, Christos Faloutsos (2010)
We use Python for programming in this project.
- Network Analysis - NetworkX: https://networkx.org/documentation/stable/reference/index.html
- Anomaly Detection - cada: https://github.com/thomashelling/cada
- Modularity Maximization via Leiden Algorithm - CDlib: https://cdlib.readthedocs.io/en/latest/index.html
See resources here.
- Network Database - Neo4j: https://neo4j.com/developer/get-started/