A Study of Anomaly Detection in Bipartite Graph

Many real applications can be modeled using bipartite graphs, such as users vs. files, traders vs. stocks, conferences vs. authors, and so on. Bipartite graph perform the operation finding similar nodes (Neighborhood Formation) and abnormal nodes (Anomaly detection). To propose algorithms to compute...

Full description

Saved in:
Bibliographic Details
Published in:International journal of advanced networking and applications 2017-01, Vol.8 (5), p.128-130
Main Author: Punitha, K
Format: Article
Language:English
Subjects:
Algorithms
Anomalies
Datasets
Graph representations
Graph theory
Linear algebra
Neighborhoods
Nodes
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Many real applications can be modeled using bipartite graphs, such as users vs. files, traders vs. stocks, conferences vs. authors, and so on. Bipartite graph perform the operation finding similar nodes (Neighborhood Formation) and abnormal nodes (Anomaly detection). To propose algorithms to compute graph partitioning and also propose algorithms to identify abnormal nodes, using normality scores (ns) based on relevance scores (rs). Evaluate the quality of the datasets, and also measure the performance of the anomaly detection algorithm with manually injected anomalies. Effectiveness and efficiency of the method are confirmed by experiments on several real datasets.
ISSN:0975-0290
0975-0282