Community-based anomaly detection in evolutionary networks

Abstract

Networks of dynamic systems, including social networks, the World Wide Web, climate networks, and biological networks, can be highly clustered. Detecting clusters, or communities, in such dynamic networks is an emerging area of research; however, less work has been done in terms of detecting community-based anomalies. While there has been some previous work on detecting anomalies in graph-based data, none of these anomaly detection approaches have considered an important property of evolutionary networks-their community structure. In this work, we present an approach to uncover community-based anomalies in evolutionary networks characterized by overlapping communities. We develop a parameter-free and scalable algorithm using a proposed representative-based technique to detect all six possible types of community-based anomalies: grown, shrunken, merged, split, born, and vanished communities. We detail the underlying theory required to guarantee the correctness of the algorithm. We measure the performance of the communitybased anomaly detection algorithm by comparison to a non-representative-based algorithm on synthetic networks, and our experiments on synthetic datasets show that our algorithm achieves a runtime speedup of 11-46 over the baseline algorithm. We have also applied our algorithm to two real-world evolutionary networks, Food Web and Enron Email. Significant and informative community-based anomaly dynamics have been detected in both cases.

A Scalable Approach for Outlier Detection in Edge Streams Using Sketch-based Approximations

“…G t is compared to G t−1 ) and a scoring function maps the graph pair to a real number, creating a time series of scores. The scoring function can be based on community detection [8,11,12], graph distance [10,17], tensor decomposition [20,21], compression [19], or other graph features. A threshold is then applied to the time series of scores to identify events (isolated abnormalities), or change points (time points where significant changes occur in the graph and then persist).…”

Hybrid Optimization Algorithm for Community and Fraud Detection in Complex Networks for High Immunity Towards Link and Node Failures

“…The temporal dynamics plays a vital role while integrating provides a better perceptive of network behavior [1][2][3]. Basically, the community relates the grouping of nodes with a cluster connected with many edges and cluster exists with few edges [4].…”

“…The networks are becoming wider and wider since it is the period of information explosion. Thus, we required many effective community detection algorithms for analyzing the networks with millions of vertices [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15].…”

Data Mining Approach in Climate Classification and Climate Network Construction – Case Study Montenegro

“…A link between two nodes exists if there is a significant statistical interdependence between their time series. Typically, the linear cross-correlation function is used as the simplest measure of the statistical interdependence of temporal series [11].…”