Inet: Visual Analysis of Irregular Transition in Multivariate Dynamic Networks
Han Dongming,Pan Jiacheng,Pan Rusheng,Zhou Dawei,Cao Nan,He Jingrui,Xu Mingliang,Chen Wei
DOI: https://doi.org/10.1007/s11704-020-0013-1
IF: 2.6688
2021-01-01
Frontiers of Computer Science
Abstract:Multivariate dynamic networks indicate networks whose topology structure and vertex attributes are evolving along time. They are common in multimedia applications. Anomaly detection is one of the essential tasks in analyzing these networks though it is not well addressed. In this paper, we combine a rare category detection method and visualization techniques to help users to identify and analyze anomalies in multivariate dynamic networks. We conclude features of rare categories and two types of anomalies of rare categories. Then we present a novel rare category detection method, called DIRAD, to detect rare category candidates with anomalies. We develop a prototype system called iNet, which integrates two major visualization components, including a glyph-based rare category identifier, which helps users to identify rare categories among detected substructures, a major view, which assists users to analyze and interpret the anomalies of rare categories in network topology and vertex attributes. Evaluations, including an algorithm performance evaluation, a case study, and a user study, are conducted to test the effectiveness of proposed methods.