DOCSLIB.ORG

Modeling and Identifying Russian Troll Accounts on Twitter

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Modeling and Identifying Russian Troll Accounts on Twitter Still Out There: Modeling and Identifying Russian Troll Accounts on Twitter Jane Im,y Eshwar Chandrasekharan,yy Jackson Sargent,y Paige Lighthammer,y Taylor Denby,y Ankit Bhargava,y Libby Hemphill,y David Jurgens,y Eric Gilberty yUniversity of Michigan yyGeorgia Institute of Technology [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected] Abstract Twitter1 and “normal”, control accounts (170K accounts, or There is evidence that Russia’s Internet Research Agency 98.6% of our sample) collected by the authors. (See Figure 1 attempted to interfere with the 2016 U.S. election by run- for a visual overview of the process used in this work.) Our ning fake accounts on Twitter—often referred to as “Russian goals are to determine whether “new” trolls can be identified trolls”. In this work, we: 1) develop machine learning models by models built on “old” trolls and to demonstrate that troll that predict whether a Twitter account is a Russian troll within detection is both possible and efficient, even with “old” data. a set of 170K control accounts; and, 2) demonstrate that it is We find that it is possible to disambiguate between a Rus- possible to use this model to find active accounts on Twitter sian troll account and a large number of these randomly se- still likely acting on behalf of the Russian state. Using both lected control accounts among users. One model, a simple behavioral and linguistic features, we show that it is possible logistic regression, achieves a precision of 78.5% and an to distinguish between a troll and a non-troll with a precision of 78.5% and an AUC of 98.9%, under cross-validation. Ap- AUC of 98.9%. Next we asked whether it was possible to plying the model to out-of-sample accounts still active today, use the model trained on past data to unmask Russian trolls we find that up to 2.6% of top journalists’ mentions are oc- currently active on Twitter (see Figure 2 for an example)? cupied by Russian trolls. These findings imply that the Rus- The logistic regression is attractive in this context as its sim- sian trolls are very likely still active today. Additional anal- plicity seems most likely to generalize to out-of-sample data. ysis shows that they are not merely software-controlled bots, Toward that end, we apply our classifier to Twitter accounts and manage their online identities in various complex ways. that mentioned high-profile journalists in late 2018. We find Finally, we argue that if it is possible to discover these ac- the computational model flags 3.7% of them as statistically counts using externally-accessible data, then the platforms— likely Russian trolls and find reasonable agreement between with access to a variety of private internal signals—should our classifier and human labelers. succeed at similar or better rates. Our model allows us to estimate the activity of trolls. As a case study, we estimate the activity of suspected Russian 1 Introduction troll accounts engaging in one type of adversarial campaign: It is widely believed that Russia’s Internet Research Agency engaging with prominent journalists. Since we have no way (IRA) tried to interfere with the 2016 U.S. election as well of truly knowing which of these model-identified accounts as other elections by running fake accounts on Twitter— are truly Russian trolls—perhaps only the IRA knows this— often called the “Russian troll” accounts (Gorodnichenko, we perform a secondary human evaluation in order to estab- Pham, and Talavera 2018; Ferrara 2017; Stella, Ferrara, lish consensus on whether the model is identifying validly and De Domenico 2018). This interference could have im- suspicious accounts. Our human evaluation process suggests mense consequences considering the viral nature of some that roughly 70% of these model-flagged accounts—all of tweets (Mustafaraj and Metaxas 2010; Metaxas and Musta- them still currently active on Twitter—are highly likely to be arXiv:1901.11162v1 [cs.SI] 31 Jan 2019 faraj 2012), the number of users exposed to Russian trolls’ Russian trolls. As a result, we estimate that Russian trolls oc- content (Isaac and Wakabayashi 2017; Spangher et al. 2018), cupy 2.6% of the mentions of high-profile journalists’ men- and the critical role social media have played in past polit- tions. Moreover, we find that in contrast with some prevail- ical campaigns (Cogburn and Espinoza-Vasquez 2011). In ing narratives surrounding the Russian troll program, the this paper, we develop models on a dataset of Russian trolls model-flagged accounts do not score highly on the well- active on Twitter during the 2016 U.S. elections to predict known Botometer scale (Davis et al. 2016), indicating that currently active Russian trolls. We construct machine learn- they are not simply automated software agents. ing classifiers using profile elements, behavioral features, Finally, we perform an exploratory open coding of the language distribution, function word usage, and linguistic identity deception strategies used by the currently active ac- features, on a highly unbalanced dataset of Russian troll ac- counts discovered by our model. For instance, some pretend counts (2.2K accounts, or 1.4% of our sample) released by Copyright c 2018, Association for the Advancement of Artificial 1https://about.twitter.com/en_us/values/ Intelligence (www.aaai.org). All rights reserved. elections-integrity.html#data Figure 1: Flowchart illustrating the steps of our research pipeline. to be an American mother or a middle-aged white man via case of the 2016 E.U. Referendum (Brexit Referendum), and profile pictures and descriptions, but their tweet rates are ab- then dropped afterwards (Gorodnichenko, Pham, and Talav- normally high, and their tweets revolve solely around politi- era 2018). Furthermore, it is suspected that Russia is be- cal topics. hind the MacronLeaks campaign that occurred during the This paper makes the following three contributions, build- 2017 French presidential elections period (Ferrara 2017), ing on an emerging line of scholarship around the Russian as well as the Catalonian referendum (Stella, Ferrara, and troll accounts (Stewart, Arif, and Starbird 2018; Spangher et De Domenico 2018). al. 2018; Griffin and Bickel 2018; Zannettou et al. 2018b; Boatwright, Linvill, and Warren 2018; Boyd et al. 2018). 2.2 Emerging Work on the 2016 Russian Trolls First, we show that it is possible to separate Russian trolls While a brand new area of scholarship, emerging work has from other accounts in the data previous to 2019, and that examined the datasets of Russian trolls released by Twitter. this computational model is still accurate on 2019 data. As a Researchers from Clemson University identified five cate- corollary, we believe this work establishes that a large num- gories of trolls and argued the behavior between these cat- ber of Russian troll accounts are likely to be currently active egories were radically different (Boatwright, Linvill, and on Twitter. Second, we provide our model to the research 2 Warren 2018). This was especially marked for left- and community. This will enable other researchers to study their right-leaning accounts (the dataset contains both). For in- own questions about the trolls, such as “What are their ob- stance, the IRA promoted more left-leaning content than jectives?” and “How are they changing over time?” Third, right-leaning on Facebook, while right-leaning Twitter han- we find that accounts flagged by our model as Russian trolls dles received more engagement. (Spangher et al. 2018). are not merely bots but use diverse ways to build and man- New work has looked at how the Russian troll accounts age their online identities. Finally, we argue that if it is pos- were retweeted in the context of the #BlackLivesMatter sible to discover these accounts using externally-accessible movement (Stewart, Arif, and Starbird 2018)—a movement data, then the social platforms—with access to a variety of targeted by the trolls. The retweets were divided among dif- private, internal signals—should succeed at similar or better ferent political perspectives and the trolls took advantage of rates at finding and deactivating Russian troll accounts. this division. There is some disagreement about how pre- dictable the Russian trolls are. Griffin and Bickel (2018) ar- 2 Related Work gue that the Russian trolls are composed of accounts with First, we review what is known about Russian’s interfer- common but customized behavioral characteristics that can ence in Western democracies via online campaigns, and then be used for future identification (Griffin and Bickel 2018), move on to the emerging work on these 2016 election related while other work has shown that the trolls’ tactics and tar- Russian trolls themselves. We conclude by discussing work gets change over time, implying that the task of automatic on social bots, and by reviewing theories of online deception detection is not simple (Zannettou et al. 2018b). Finally, the that inform the quantitative approaches in this paper. Russian trolls show unique linguistic behavior as compared to a baseline cohort (Boyd et al. 2018). 2.1 Russia’s Interference on Political Campaigns Users Who Interact with the Trolls. Recent work has While state-level online interference in democratic pro- also examined the users who interact with the Russian trolls cesses is an emerging phenomenon, new research docu- on Twitter. For example, misinformation produced by the ments Russia’s online political manipulation campaigns in Russian trolls was shared more often by conservatives than countries other than the United States. For instance, previ- liberals on Twitter (Badawy, Ferrara, and Lerman 2018). ous work has shown that a high volume of Russian tweets Models can predict which users will spread the trolls’ con- were generated a few days before the voting day in the tent by making use of political ideology, bot likelihood scores, and activity-related account metadata (Badawy, Ler- 2URL available after blind review.
Load more

Recommended publications
  • Automated Tackling of Disinformation
    Automated tackling of disinformation STUDY Panel for the Future of Science and Technology European Science-Media Hub EPRS | European Parliamentary Research Service Scientific Foresight Unit (STOA) PE 624.278 – March 2019 EN Automated tackling of disinformation Major challenges ahead This study maps and analyses current and future threats from online misinformation, alongside currently adopted socio-technical and legal approaches. The challenges of evaluating their effectiveness and practical adoption are also discussed. Drawing on and complementing existing literature, the study summarises and analyses the findings of relevant journalistic and scientific studies and policy reports in relation to detecting, containing and countering online disinformation and propaganda campaigns. It traces recent developments and trends and identifies significant new or emerging challenges. It also addresses potential policy implications for the EU of current socio-technical solutions. ESMH | European Science-Media Hub AUTHORS This study was written by Alexandre Alaphilippe, Alexis Gizikis and Clara Hanot of EU DisinfoLab, and Kalina Bontcheva of The University of Sheffield, at the request of the Panel for the Future of Science and Technology (STOA). It has been financed under the European Science and Media Hub budget and managed by the Scientific Foresight Unit within the Directorate-General for Parliamentary Research Services (EPRS) of the Secretariat of the European Parliament. Acknowledgements The authors wish to thank all respondents to the online survey, as well as first draft, WeVerify, InVID, PHEME, REVEAL, and all other initiatives that contributed materials to the study. ADMINISTRATOR RESPONSIBLE Mihalis Kritikos, Scientific Foresight Unit To contact the publisher, please e-mail [email protected] LINGUISTIC VERSION Original: EN Manuscript completed in March 2019.
    [Show full text]
  • Recent Trends in Online Foreign Influence Efforts
    Recent Trends in Online Foreign Influence Efforts Diego A. Martin, Jacob N. Shapiro, Michelle Nedashkovskaya Woodrow Wilson School of Public and International Affairs Princeton University Princeton, New Jersey, United States E-mail: [email protected] Email: [email protected] E-mail: [email protected] Abstract: Foreign governments have used social media to influence politics in a range of countries by promoting propaganda, advocating controversial viewpoints, and spreading disinformation. We analyze 53 distinct foreign influence efforts (FIEs) targeting 24 different countries from 2013 through 2018. FIEs are defined as (i) coordinated campaigns by one state to impact one or more specific aspects of politics in another state (ii) through media channels, including social media, (iii) by producing content designed to appear indigenous to the target state. The objective of such campaigns can be quite broad and to date have included influencing political decisions by shaping election outcomes at various levels, shifting the political agenda on topics ranging from health to security, and encouraging political polarization. We draw on more than 460 media reports to identify FIEs, track their progress, and classify their features. Introduction Information and Communications Technologies (ICTs) have changed the way people communi- cate about politics and access information on a wide range of topics (Foley 2004, Chigona et al. 2009). Social media in particular has transformed communication between leaders and voters by enabling direct politician-to-voter engagement outside traditional avenues, such as speeches and press conferences (Ott 2017). In the 2016 U.S. presidential election, for example, social media platforms were more widely viewed than traditional editorial media and were central to the campaigns of both Democratic candidate Hillary Clinton and Republican candidate Donald Trump (Enli 2017).
    [Show full text]
  • Empirical Comparative Analysis of Bot Evidence in Social Networks
    Bots in Nets: Empirical Comparative Analysis of Bot Evidence in Social Networks Ross Schuchard1(&) , Andrew Crooks1,2 , Anthony Stefanidis2,3 , and Arie Croitoru2 1 Computational Social Science Program, Department of Computational and Data Sciences, George Mason University, Fairfax, VA 22030, USA [email protected] 2 Department of Geography and Geoinformation Science, George Mason University, Fairfax, VA 22030, USA 3 Criminal Investigations and Network Analysis Center, George Mason University, Fairfax, VA 22030, USA Abstract. The emergence of social bots within online social networks (OSNs) to diffuse information at scale has given rise to many efforts to detect them. While methodologies employed to detect the evolving sophistication of bots continue to improve, much work can be done to characterize the impact of bots on communication networks. In this study, we present a framework to describe the pervasiveness and relative importance of participants recognized as bots in various OSN conversations. Specifically, we harvested over 30 million tweets from three major global events in 2016 (the U.S. Presidential Election, the Ukrainian Conflict and Turkish Political Censorship) and compared the con- versational patterns of bots and humans within each event. We further examined the social network structure of each conversation to determine if bots exhibited any particular network influence, while also determining bot participation in key emergent network communities. The results showed that although participants recognized as social bots comprised only 0.28% of all OSN users in this study, they accounted for a significantly large portion of prominent centrality rankings across the three conversations. This includes the identification of individual bots as top-10 influencer nodes out of a total corpus consisting of more than 2.8 million nodes.
    [Show full text]
  • Hacks, Leaks and Disruptions | Russian Cyber Strategies
    CHAILLOT PAPER Nº 148 — October 2018 Hacks, leaks and disruptions Russian cyber strategies EDITED BY Nicu Popescu and Stanislav Secrieru WITH CONTRIBUTIONS FROM Siim Alatalu, Irina Borogan, Elena Chernenko, Sven Herpig, Oscar Jonsson, Xymena Kurowska, Jarno Limnell, Patryk Pawlak, Piret Pernik, Thomas Reinhold, Anatoly Reshetnikov, Andrei Soldatov and Jean-Baptiste Jeangène Vilmer Chaillot Papers HACKS, LEAKS AND DISRUPTIONS RUSSIAN CYBER STRATEGIES Edited by Nicu Popescu and Stanislav Secrieru CHAILLOT PAPERS October 2018 148 Disclaimer The views expressed in this Chaillot Paper are solely those of the authors and do not necessarily reflect the views of the Institute or of the European Union. European Union Institute for Security Studies Paris Director: Gustav Lindstrom © EU Institute for Security Studies, 2018. Reproduction is authorised, provided prior permission is sought from the Institute and the source is acknowledged, save where otherwise stated. Contents Executive summary 5 Introduction: Russia’s cyber prowess – where, how and what for? 9 Nicu Popescu and Stanislav Secrieru Russia’s cyber posture Russia’s approach to cyber: the best defence is a good offence 15 1 Andrei Soldatov and Irina Borogan Russia’s trolling complex at home and abroad 25 2 Xymena Kurowska and Anatoly Reshetnikov Spotting the bear: credible attribution and Russian 3 operations in cyberspace 33 Sven Herpig and Thomas Reinhold Russia’s cyber diplomacy 43 4 Elena Chernenko Case studies of Russian cyberattacks The early days of cyberattacks: 5 the cases of Estonia,
    [Show full text]
  • Artificial Intelligence and Countering Violent Extremism: a Primer
    Artificial Intelligence and Countering Violent Extremism: A Primer ByMarie Hxxnry Schroeter Mxxhm, Inxxs Oxxlxxmxxnxx, xxnd Fxxnxx Sxxngxxr GNET xxsis a xxspecial spxxcxxl project prxxjxxct delivered dxxlxxvxxrxxd by the International by thxx Intxxrnxxtxxnxxl Centre Cxxntrxx fxxrfor the thxx Study Stxxdy of Radicalisation,xxf Rxxdxxcxxlxxsxxtxxn, King’s College Kxxng’s London. Cxxllxxgxx Lxxndxxn. The author of this report is Marie Schroeter, Mercator Fellow on New Technology in International Relations: Potentials and Limitations of Artifical Intelligence to Prevent Violent Extremism Online The Global Network on Extremism and Technology (GNET) is an academic research initiative backed by the Global Internet Forum to Counter Terrorism (GIFCT), an independent but industry‑funded initiative for better understanding, and counteracting, terrorist use of technology. GNET is convened and led by the International Centre for the Study of Radicalisation (ICSR), an academic research centre based within the Department of War Studies at King’s College London. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing those, either expressed or implied, of GIFCT, GNET or ICSR. We would like to thank Tech Against Terrorism for their support with this report. CONTACT DETAILS For questions, queries and additional copies of this report, please contact: ICSR King’s College London Strand London WC2R 2LS United Kingdom T. +44 20 7848 2098 E. [email protected] Twitter: @GNET_research Like all other GNET publications, this report can be downloaded free of charge from the GNET website at www.gnet‑research.org. © GNET Artificial Intelligence and Countering Violent Extremism: A Primer Executive Summary Radicalisation can take place offline as well as online.
    [Show full text]
  • Developing a Twitter Bot That Can Join a Discussion Using State-Of-The-Art Architectures
    DEVELOPING A TWITTER BOT THAT CAN JOIN A DISCUSSION USING STATE-OF-THE-ART ARCHITECTURES A THESIS SUBMITTED TO THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES OF MIDDLE EAST TECHNICAL UNIVERSITY BY YUSUF MÜCAHIT˙ ÇETINKAYA˙ IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN COMPUTER ENGINEERING AUGUST 2019 Approval of the thesis: DEVELOPING A TWITTER BOT THAT CAN JOIN A DISCUSSION USING STATE-OF-THE-ART ARCHITECTURES submitted by YUSUF MÜCAHIT˙ ÇETINKAYA˙ in partial fulfillment of the require- ments for the degree of Master of Science in Computer Engineering Department, Middle East Technical University by, Prof. Dr. Halil Kalıpçılar Dean, Graduate School of Natural and Applied Sciences Prof. Dr. Mehmet Halit Oguztüzün˘ Head of Department, Computer Engineering Prof. Dr. Ismail˙ Hakkı Toroslu Supervisor, Computer Engineering, METU Examining Committee Members: Prof. Dr. Tolga Can Computer Engineering, METU Prof. Dr. Ismail˙ Hakkı Toroslu Computer Engineering, METU Prof. Dr. Ahmet Co¸sar Computer Engineering, UTAA Date: I hereby declare that all information in this document has been obtained and presented in accordance with academic rules and ethical conduct. I also declare that, as required by these rules and conduct, I have fully cited and referenced all material and results that are not original to this work. Name, Surname: Yusuf Mücahit Çetinkaya Signature : iv ABSTRACT DEVELOPING A TWITTER BOT THAT CAN JOIN A DISCUSSION USING STATE-OF-THE-ART ARCHITECTURES Çetinkaya, Yusuf Mücahit M.S., Department of Computer Engineering Supervisor: Prof. Dr. Ismail˙ Hakkı Toroslu August 2019, 118 pages Twitter is today mostly used for sharing and commenting about news [1].
    [Show full text]
  • WEAPONS of MASS DISTRACTION: Foreign State-Sponsored Disinformation in the Digital Age
    WEAPONS OF MASS DISTRACTION: Foreign State-Sponsored Disinformation in the Digital Age MARCH 2019 PARK ADVISORS | Weapons of Mass Distraction: Foreign State-Sponsored Disinformation in the Digital Age Authored by Christina Nemr and William Gangware Acknowledgements The authors are grateful to the following subject matter experts who provided input on early drafts of select excerpts: Dr. Drew Conway, Dr. Arie Kruglanski, Sean Murphy, Dr. Alina Polyakova, and Katerina Sedova. The authors also appreciate the contributions to this paper by Andrew Rothgaber and Brendan O’Donoghue of Park Advisors, as well as the editorial assistance provided by Rhonda Shore and Ryan Jacobs. This report was produced with support from the US Department of State’s Global Engagement Center. Any views expressed in this report are those of the authors and do not necessarily reflect the views of the US State Department, Park Advisors, or its subject matter expert consultants. Any errors contained in this report are the authors’ alone. PARK ADVISORS | Weapons of Mass Distraction: Foreign State-Sponsored Disinformation in the Digital Age 0. Table of Contents 01 Introduction and contextual analysis 04 How do we define disinformation? 06 What psychological factors drive vulnerabilities to disinformation and propaganda? 14 A look at foreign state-sponsored disinformation and propaganda 26 Platform-specific challenges and efforts to counter disinformation 39 Knowledge gaps and future technology challenges PARK ADVISORS | Weapons of Mass Distraction: Foreign State-Sponsored Disinformation in the Digital Age 1 Introduction and 1. contextual analysis On July 12, 2014, viewers of Russia’s main state-run television station, Channel One, were shown a horrific story.
    [Show full text]
  • 2021-CICS-Horizon-Report Master.Pdf
    LEADERSHIP TEAM Dr. Dennis A. Trinkle, Director, Center for Information and Communication Sciences Dennis Trinkle is the Director for the Center of Information and Communication Sciences (CICS) and Director of the Applied Research Institutes at Ball State University. Dr. Trinkle brings a diverse multi-sector leadership background to his leadership of CICS, where he also holds the rank of Professor of Information Sciences and Communication. Prior to joining Ball State, Dr. Trinkle served as the system chief executive, provost, and chief academic officer for Harrison College. Dr. Trinkle has served leading institutions across the higher education, corporate, and government sectors, including serving as the chief executive officer of IHETS, chief information officer, associate vice president for academic affairs, and Tenzer Professor of Infor- mation Technology for DePauw University, and president, chief executive officer, and board chair for many companies and non-profit organizations. He is the author and editor of 17 books and more than 50 articles on entrepreneurship, technology, leadership, teaching and learning, and history. Together with Kevin Allen, CEO of EiGames, he developed the award-winning video game on the entrepreneurial process—Fireworks: The Entrepreneurship Experience (2018). His latest book: Recipes for Success: How to Start a Business was published in spring 2021. Christopher Nouhan, Graduate Student, Center for Information and Communication Sciences Christopher Nouhan is a Graduate Assistant for the Center for Information and Communication Sciences (CICS) at Ball State University. Prior to enrolling in CICS, Christopher attended Alma College, where he obtained a B.A. in New Media Studies. While at Alma College, he founded Bitworks Productions, a pre-pro- fessional program that bridged media students with community clients, providing professional development opportunities to his peers.
    [Show full text]
  • World Changing Ideas for 2018
    INSIDE NIKE’S DIGITAL PLAYBOOK / NINTENDO SWITCHES IT UP, AGAIN WORLD CHANGING IDEAS TWITTER FOR 2018 L’Oréal, The North Face, FACES L.A. Mayor Eric Garcetti, THE and more TRUTH The Creative HOW JACK DORSEY’S Minds Behind FREE-SPEECH PLATFORM Casper,Allbirds WAS HIJACKED & Birchbox BY THE DARK SIDE BY AUSTIN CARR AND HARRY McCRACKEN We believe that there is no mold for invention. The proof? The Genesis G80 Sport. With its impeccably crafted balance of power and detail, aggressive dark-chrome crosshatched front grille, carbon-fiber-trimmed dash, and 19-inch copper accented alloy wheels artfully surrounding a 365-hp twin-turbo engine, it’s sure to pin you to its copper stitched sport seats. Add in the peace of mind of our most advanced safety features standard and Complimentary Service Valet, and the only thing the G80 Sport doesn’t have is limits. GENESIS.COM Optional features shown. 365-hp rating with premium fuel only. Service Valet available from Authorized Genesis Retailers during complimentary maintenance period. Includes normal wear, parts replacement and warranty repairs for 3 years or 36,000 miles, whichever comes first. Coverage area varies by retailer. Charges and fees may apply. Applies to original owner/lessee only. AS A RULE, CHALLENGE THE RULE. WON’T PAY FOR ITSELF. Switch to GEICO and save money for the things you love. Maybe it’s that high-end client dinner. Or the last-minute flight overseas. Doing business is what you love – and it doesn’t come cheap. So switch to GEICO, because you could save 15% or more on car insurance.
    [Show full text]
  • Trends in Online Influence Efforts
    Trends in Online Influence Efforts∗ Diego A. Martiny Jacob N. Shapiroz Julia G. Ilhardtx Version 2.0 August 5, 2020 Abstract Information and Communications Technologies (ICTs) create novel opportuni- ties for a wide range of political actors. Foreign governments have used social media to influence politics in a range of countries by promoting propaganda, advocating controversial viewpoints, and spreading disinformation. This report updates previ- ous work with data on 76 such foreign influence efforts (FIE) targeting 30 different countries from 2013 through 2019, as well as 20 domestic influence efforts (DIE) in which governments targeted their own citizens. Influence efforts (IEs) are de- fined as: (i) coordinated campaigns by a state or the ruling party in an autocracy to impact one or more specific aspects of politics at home or in another state, (ii) through media channels, including social media, by (iii) producing content designed to appear indigenous to the target state. The objective of such campaigns can be quite broad and to date have included shaping election outcomes at various levels, shifting the political agenda on topics ranging from health to security, and encour- aging political polarization. Our data draw on more than 920 media reports and 380 research articles/reports to identify IEs, track their progress, and classify their features. ∗We are grateful to a range of colleagues including Laura Courchesne, Nick Feamster, Andy Guess, Hans Klein, Brendan Stewart, and Alicia Wanless for helpful comments and feedback. Will Lowe provided invaluable advice on data structure and data entry. Jordan Allen, Nicola Bariletto, Arya Goel, Danielle Hull, Janette Lu, Imane Mabrouk, Matthew Merrigan, Justinas Mickus, Brendan O'Hara, Jan Oledan, Kamila Radjabova, Nina Sheridan, Joe Shipley, Jack Tait, Kamya Yadav, and Luca Zanotti provided excellent research assistance.
    [Show full text]
  • "Deep Faking" Political Twitter Using Transfer Learning and GPT-2
    "Deep Faking" Political Twitter using Transfer learning and GPT-2 Ryan Ressmeyer Sam Masling Department of Electrical Engineering Department of Computer Science Stanford University Stanford University [email protected] [email protected] Madeline Liao Department of Computer Science Stanford University [email protected] Abstract In the modern political climate, Twitter has become one of, if not the most impactful mediums for both current and aspiring politicians to communicate with their constituents. Our group attempted to replicate a variety of politicians’ twitter accounts using the recently released GPT-2 model from OpenAI and transfer learning. We compared our results to former state-of-the-art LSTM recurrent models to illustrate the advantages of transfer learning. Additionally, we used GPT-2 to extrapolate from a given context, allowing for robust tweet mapping from one user to another. 1 Introduction In recent election years, Twitter has become an increasingly important communication platform for American politicians. Our goal was to see if we could use state-of-the-art NLP techniques to not only recognize the different style and contents of tweets written by different politicians, but to actually generate similar tweets. This idea stemmed from two popular fields of study within NLP: Twitter sentiment analysis, and zero-shot text generation. By utilizing transfer learning and building upon OpenAI’s GPT-2 text generation model, we sought to train a model for the domain-specific task of generating tweets based on a given account. OpenAI’s GPT-2 is a transformer-based language model trained with the goal of predicting the next word given all previous words of some text [8].
    [Show full text]
  • A Comparative Study of Bot Detection Techniques Methods with an Application Related to Covid-19 Discourse on Twitter
    A comparative study of Bot Detection techniques methods with an application related to Covid-19 discourse on Twitter. Marzia Antenore∗1, Jos´eM. Camacho-Rodr´ıguez †2, Emanuele Panizzi‡2 1Department of Communication and Social Research, Sapienza University of Rome 2Department of Computer Science, Sapienza University of Rome Abstract Bot Detection is an essential asset in a period where Online Social Networks(OSN) is a part of our lives. This task becomes more relevant in crises, as the Covid-19 pandemic, where there is an incipient risk of proliferation of social bots, producing a possible source of misinformation. In order to address this issue, it has been compared different methods to detect automatically social bots on Twitter using Data Selection. The techniques utilized to elaborate the bot detection models include the utilization of features as the tweets' metadata or the Digital Fingerprint of the Twitter accounts. In addition, it was analyzed the presence of bots in tweets from different periods of the first months of the Covid-19 pandemic, using the bot detection technique which best fits the scope of the task. Moreover, this work includes also analysis over aspects regarding the discourse of bots and humans, such as sentiment or hashtag utilization. arXiv:2102.01148v1 [cs.SI] 1 Feb 2021 1 Introduction According to [47], a bot is a socio-technical entity based on a software program whose aim is to simulate human behavior in Online Social Networks(OSN) such as Facebook, Twitter, or Instagram. Bots are configured to resemblance as humans not just to other human users, but also to the OSN platform [28].
    [Show full text]