**9. ConclusionS**

In this paper, we propose a novel visualization system, which has novel visual glyphs and uses multi-view to explore, detect and analyze the anomaly in the social network. The system analyzes from macroscopic, mesoscopic and microscopic perspectives. We show the abnormal situation in the online social communication network after anomaly detection from a macroscopic point of view; in the mesoscopic view, we introduce galaxy maps, combined with the ego central network analysis method, to display the interested users in multi-dimensional, from the network structure, active time, alters intimacy and other aspects to judge the abnormal degree of users; and through the microscopic view, combined with timing, we can evaluate the abnormal degree of users from the point of view of alters. We also add friendly and intuitive interactions to help researchers quickly ge<sup>t</sup> the information they want. We use a call record data to demonstrate the system is beneficial for detecting abnormal behavior in online social communication. We also discuss the feasibility of applying the method to other fields, like IOT and cyber-physical social systems.

However, limited by time and energy, our work still has a lot of room to improve. Through the case study, we find that although the LOF algorithm can help us to mine latent anomalous egos by combining time series, it also incorrectly classifies some normal errors. Restricted by datasets, it is difficult for us to analyze alien alters and egos, which is disadvantageous to our analysis.

In the future, we plan to design better anomaly detection algorithms. This can make our detection accuracy higher. Besides, the dataset used in this experiment is only provided by a certain operator, so there are limitations in the analysis of specific contacts. In the follow-up experiments, we hope to deepen cooperation with other operators, obtain more and more communication data from the external network, and conduct more in-depth research.

**Author Contributions:** Conceptualization, J.P.; methodology, J.P. and J.Z.; software, J.Z. and H.S.; validation, J.P. and J.Z.; formal analysis, J.P. and J.Z. and H.S. and T.Z.; resources, J.P.; data curation, J.Z. and T.Z.; writing—original draft preparation, J.P. and J.Z.; writing—review and editing, J.P. and J.Z. and Y.R.; visualization, J.Z. and H.S.; supervision, J.P. and Y.R.; project administration, J.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China, gran<sup>t</sup> number 61872066 and U19A2078 and the Science and Technology project of Sichuan, number 2020YFG0056, 2019YFG0504 and 2020YFG0459 and the Science and Technology Service Industry Demonstration project of Sichuan, number 2019GFW126 and the Aeronautic Science Foundation of China, number 20160580004.

**Conflicts of Interest:** The authors declare no conflict of interest.
