Research Article
CSCD: A Cyber Security Community Detection Scheme on Online Social Networks
@INPROCEEDINGS{10.1007/978-3-031-36574-4_8, author={Yutong Zeng and Honghao Yu and Tiejun Wu and Yong Chen and Xing Lan and Cheng Huang}, title={CSCD: A Cyber Security Community Detection Scheme on Online Social Networks}, proceedings={Digital Forensics and Cyber Crime. 13th EAI International Conference, ICDF2C 2022, Boston, MA, November 16-18, 2022, Proceedings}, proceedings_a={ICDF2C}, year={2023}, month={7}, keywords={online social network community detection cyber security social network analysis}, doi={10.1007/978-3-031-36574-4_8} }- Yutong Zeng
Honghao Yu
Tiejun Wu
Yong Chen
Xing Lan
Cheng Huang
Year: 2023
CSCD: A Cyber Security Community Detection Scheme on Online Social Networks
ICDF2C
Springer
DOI: 10.1007/978-3-031-36574-4_8
Abstract
Online social networks (OSNs) are playing a crucial role in daily life, cyber security guys such as hackers, cyber criminals, and researchers also like to communication and publish opinions. Their discussions and relations can provide unprecedented opportunities for researcher to develop better insights about those accounts’ activities in communities, which could be helpful for different purposes like cyber threat intelligent hunting and attack attribution. In this paper, we propose a scheme for cyber security community detection named CSCD on OSNs. We present a social relevance analysis method by building an ego network from one seed account. Through multidimensional analysis, features organized into four categories are taken into consideration and a recognition model is used to detect security-related accounts. Then we construct the social network, consisting of detected accounts, and propound a pruning strategy to remove weak relationships between accounts on the basis of edge features. An unsupervised overlapping community detection model is applied to unearthing potential communities. To evaluate our proposed scheme, we utilize Twitter as the platform to construct datasets. The recognition model achieves an accuracy up to 95.1%, and the community detection model obtains the best performance comparing to other former algorithms.
