A Content-Based Method for Sybil Detection in Online Social Networks via Deep Learning

Abstract

Online social networks (OSNs) are generally susceptible to Sybil attack, which causes a series of cybersecurity problems and privacy violations. Malicious attackers can create massive Sybils and further utilize those fake identities to launch various Sybil attacks. Therefore, Sybil detection in OSNs has become an urgent security research problem for both academia and industries. The existing content-based methods to detect Sybils base on the combination of manual-design features and machine learning algorithms, which requires lots of professional experiences and human effort. These methods divide the Sybil detection problem into two piece-wise sub-problems, which prevents us from getting the optimal solution. In this work, we propose a novel content-based method to detect Sybils. The proposed method is an end-to-end classification model that extracts features directly from the input data, and then output the classification results in a unified framework. The proposed method includes three main parts: first, the self-normalizing convolutional neural network (CNN) is adopted to extract lower features from the multi-dimensional input data; second, the bidirectional self-normalizing LSTM network (bi-SN-LSTM) is developed to extract higher features from the compressed feature map sequence; third, the dense layer and softmax classifier are stacked to output the classification results. Unlike the traditional bidirectional long short-term memory network (bi-LSTM), the proposed bi-SN-LSTM network utilizes SELU as the activation function of its recurrent step, which provides unbounded changes to the state value. Through the case study of the real-world dataset, the comparison experiments demonstrate that our method significantly outperforms other state-of-the-art methods. INDEX TERMS Convolutional neural network (CNN), deep learning (DL), long short-term memory (LSTM), online social networks (OSNs), sybil detection.

Social media bot detection with deep learning methods: a systematic review

“…In addition, new type of features should be investigated to boost up the accuracy for real-world implementation and ensure generalizability. Graph neural network-based methods should be explored more in addition to individual feature-based model, to facilitate the detection of social media bot cluster or coordinated attack or campaigns [45].…”

“…In a less complicated way, another study [44] showed that bot accounts could be detected with a very high accuracy with a combination of CNN and ANN using a single post on the social media; however, it requires a huge amount of tweets (about 1 million) in each class. Gao et al [45] used a self-normalizing CNN, adopted to extract lower features from the multi-dimensional input data, that Neural Computing and Applications were generated from user profiles and the structure of users' local subgraph and a bidirectional self-normalizing LSTM network (bi-SN-LSTM) to extract higher features from the compressed feature map sequence. Their unique technical contribution lied in proposing bi-SN-LSTM network with SELU as the activation function of its recurrent step, which provided unbounded changes to the state value and using alpha dropout to prevent overfitting while training neural networks.…”

A comprehensive survey on online social networks security and privacy issues: Threats, machine learning‐based solutions, and open challenges

“…Gao et al 135 proposed a content‐based Sybil detection method, where the problem of Sybil detection is divided into two subproblems. The method will do the feature extraction from the input data and perform an end‐to‐end classification and returns the list of Sybil account as output.…”

An Enhanced Sybil Guard to Detect Bots in Online Social Networks

“…However, like all graph-based techniques, the above method fails if the attacker can craftily achieve a vast network with a bunch of real users. Gao et al (2020) [7] incorporated an ensemble of deep learning algorithms which comprised of CNN to extract low level features and bidirectional LSTM to extricate correlated features. These features were fed into Softmax classifier to classify if the account is sybil or not.…”