Social Network Based Anomaly Detection of Organizational Behavior using Temporal Pattern Mining

The interaction between members within an organization evolves continuously and drives the organizational behavior to change over time. To understand the temporal behaviors of the organization and recognize the evolving characteristics of the networks, we need effective tools that can capture evolution of the objects. In this paper, we propose a novel and important problem in evolution behavior discovery. Given a series of snapshots of a temporal dataset, which are modeled as evolving social networks, our goal is to find networks which evolve in a dramatically different way that deviate from temporal pattern norm. We define such objects as organizational behavior anomalies. It is a challenging problem as temporal patterns are hidden deeply in noisy evolving datasets and thus it is difficult to distinguish anomalous objects from normal ones. To this end, we propose a social network based temporal pattern mining framework to detect organizational behavior anomalies. Specifically, we first give the definition and a brief description of organizational behavior anomalies. Then, we extract a number of features in the temporal dataset for profiling organizational networks. Third, we use the temporal pattern mining framework to evaluate anomalies of networks deviating from the normal evolutionary patterns. Within the proposed framework, a two-step procedure that exploits the strengths of unsupervised anomaly detection and supervised classification is used. Experimental results on two real-world datasets show that the proposed framework is highly effective in discovering interesting organizational behavior anomalies.