Social fingerprinting: detection of spambot groups through DNA-inspired behavioral modeling

Abstract

Spambot detection in online social networks is a long-lasting challenge involving the study and design of detection techniques capable of efficiently identifying ever-evolving spammers. Recently, a new wave of social spambots has emerged, with advanced human-like characteristics that allow them to go undetected even by current state-of-the-art algorithms. In this paper, we show that efficient spambots detection can be achieved via an in-depth analysis of their collective behaviors exploiting the digital DNA technique for modeling the behaviors of social network users. Inspired by its biological counterpart, in the digital DNA representation the behavioral lifetime of a digital account is encoded in a sequence of characters. Then, we define a similarity measure for such digital DNA sequences. We build upon digital DNA and the similarity between groups of users to characterize both genuine accounts and spambots. Leveraging such characterization, we design the Social Fingerprinting technique, which is able to discriminate among spambots and genuine accounts in both a supervised and an unsupervised fashion. We finally evaluate the effectiveness of Social Fingerprinting and we compare it with three state-of-the-art detection algorithms. Among the peculiarities of our approach is the possibility to apply off-the-shelf DNA analysis techniques to study online users behaviors and to efficiently rely on a limited number of lightweight account characteristics.

Detection of malicious social bots: A survey and a refined taxonomy

“…GroupFound achieved a detection rate of 86:27% with a false positive rate of 8:54%. Cresci et al (2018a) observed that it is not enough to merely depend on the history of previous behaviour records to detect new generations of spam bots. Instead, we need to investigate collective behaviors of users' groups to determine whether these account are bots or not.…”

“…Interestingly, the very low recall of Yang et al (2013) can be seen as an evidence of a new generation of social bots that are hard to detect when they are considered individually even using current state-of-the-art algorithms. Cresci et al (2018a) approach is flexible in terms of not focusing on specific characteristics. Moreover, it reduces the cost for data gathering by not considering the properties of the social graph.…”

VASSL: A Visual Analytics Toolkit for Social Spambot Labeling

“…R1 Show similarities among accounts. This requirement is essential for enabling users to explore different characteristics to cluster the accounts based on [4,10,11].…”

“…The system should enable users to highlight and cluster accounts interactively. This interactive clustering is essential to reveal potential group-based spamming activities [10,11].…”

User Behavior to Identify Malicious Activities in Large-Scale Social Networks

“…In other words, users form connections with the elements as they interact in the network, e.g.,through communication with other users. Researchers in this category use data and machine-learning methods to find users that do not conform to certain rules; such techniques usually compare the user activities to a predefined set of activities [1], [11], [23], [8]. In most cases, history-based algorithms need to have a set of predefined standards that describe legitimate users, hence ensuring that the systems adopt the strategy of finding Sybil profiles.…”