Malicious Account Detection based on intelligent algorithm Research

Abstract

Malicious account detection has always been a hot issue. This paper mainly identifies the malicious account at the registration level. After feature extraction of the collected data, a weighted undirected graph is constructed. Node2Vec is used to convert each node into a multidimensional vector. K-means algorithms and DBSCAN algorithms are used to obtain the model of garbage account. Finally, Euclidean distance is used to identify whether the malicious account is.