A survey on adversarial attacks on community detection in complex networks

With the rapid development of complex network analysis technology, community detection technology has also caused privacy leakage risks while mining network structure characteristics. Therefore, adversarial attack technology for community detection has emerged, which aims to hide the real community through network perturbations under a limited budget so as to protect data security. This paper first introduces the background of adversarial attack for community detection, proposes a unified modeling framework for community hiding problems, and summarizes commonly used dataset information. Subsequently, the metrics for evaluating the effect of community hiding are reviewed, and the existing methods are innovatively divided into three categories: global community hiding, target community hiding, and flexible community hiding. The core ideas, technical advantages, and limitations of different algorithms are analyzed and summarized. Finally, based on our own research foundation in community hiding, the research motivation, difficult issues, and future directions in this field are explored and discussed.