Detecting coordinated online behaviour — A multiplex network approach

Thumbnail Image

Files

master_Iannucci_Letizia_2023.pdf (4.6 MB)

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2023-12-11

Department

Major/Subject

Machine Learning, Data Science and Artificial Intelligence

Mcode

SCI3044

Degree programme

Master’s Programme in Computer, Communication and Information Sciences

Language

en

Pages

70

Series

Abstract

Given the impact of online content on individuals’ opinions and attitudes, early detection of online coordinated activities plays a vital role in mitigating the influence of disinformation and information manipulation. Recent research demonstrates that information operations and astroturfing campaigns exhibit distinguishing characteristics related to the temporal proximity and similarity of actions performed by coordinating individuals. This Thesis extends existing methods by introducing a multiplex network approach and by enhancing Newman collaboration model with a temporal dimension to represent latent coordination. The evaluation of the proposed approach on a range of simulated campaigns demonstrates that detectability significantly improves when considering multiple layers or dimensions of coordination and when modeling the significance of the inter-activity times between users. Notably, both recall and F1 metrics exhibit substantial improvements compared to monoplex approaches found in the current literature. In particular, based on the F1 scores, the proposed approach is able to outperform traditional methods that rely solely on counting co-occurrences without considering the temporal dimension. Even more crucially, the proposed time-aware model can achieve high recall even on sophisticated coordinated operations, surpassing methods that employ fixed-size time windows. This is especially significant for campaigns that span extended time intervals, up to several days, where malicious accounts strategically alternate between activities and pauses. These findings not only contribute to a deeper understanding of the challenges in coordination detection but also hold significant implications for the creation of more effective tools and strategies to safeguard the integrity of online discourse.

Description

Supervisor

Kivelä, Mikko

Thesis advisor

Matakos, Antonis

Keywords

coordination detection, community detection, multiplex networks, social networks, information operations

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202312187365

Collections

[dipl] Perustieteiden korkeakoulu / SCI