Conference PaperPDF Available

‘Conspiracy Machines’ - The Role of Social Bots during the COVID-19 ‘Infodemic’


Abstract and Figures

The omnipresent COVID-19 pandemic gave rise to a parallel spreading of misinformation, also referred to as an ‘Infodemic’. Consequently, social media have become targets for the application of social bots, that is, algorithms that mimic human behaviour. Their ability to exert influence on social media can be exploited by amplifying misinformation, rumours, or conspiracy theories which might be harmful to society and the mastery of the pandemic. By applying social bot detection and content analysis techniques, this study aims to determine the extent to which social bots interfere with COVID-19 discussions on Twitter. A total of 78 presumptive bots were detected within a sample of 542,345 users. The analysis revealed that bot-like users who disseminate misinformation, at the same time, intersperse news from renowned sources. The findings of this research provide implications for improved bot detection and managing potential threats through social bots during ongoing and future crises.
Content may be subject to copyright.
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
‘Conspiracy Machines’ - The Role of Social Bots during the
COVID-19 ‘Infodemic’
Julian Marx
Department of Computer Science and Applied Cognitive Science
University of Duisburg-Essen
Duisburg, Germany
Felix Brünker
Department of Computer Science and Applied Cognitive Science
University of Duisburg-Essen
Duisburg, Germany
Milad Mirbabaie
Faculty of Business Studies and Economics
University of Bremen
Bremen, Germany
Eric Hochstrate
Department of Computer Science and Applied Cognitive Science
University of Duisburg-Essen
Duisburg, Germany
The omnipresent COVID-19 pandemic gave rise to a parallel spreading of misinformation, also
referred to as an ‘Infodemic’. Consequently, social media have become targets for the application of
social bots, that is, algorithms that mimic human behaviour. Their ability to exert influence on social
media can be exploited by amplifying misinformation, rumours, or conspiracy theories which might be
harmful to society and the mastery of the pandemic. By applying social bot detection and content
analysis techniques, this study aims to determine the extent to which social bots interfere with COVID-
19 discussions on Twitter. A total of 78 presumptive bots were detected within a sample of 542,345
users. The analysis revealed that bot-like users who disseminate misinformation, at the same time,
intersperse news from renowned sources. The findings of this research provide implications for
improved bot detection and managing potential threats through social bots during ongoing and future
Keywords COVID-19, Social Bots, Crisis Communication, Misinformation, Infodemics.
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
1 Introduction
Social media have become a powerful communication tool for the procurement and dissemination of
information (Varol et al. 2017). Especially during extreme events, social media are used to engage with
online contacts and to stay informed (Mirbabaie et al. 2020; Mirbabaie and Marx 2019). The same
applies to the infectious Coronavirus disease 2019 (COVID-19), which turned out to be one of the most
discussed events on social media in recent years. However, apart from human users, social bots, that
is, algorithms programmed to mimic human behaviour on social media platforms, also have been
found engage in online discussions (Stieglitz, Brachten, et al. 2017). Scholarship distinguishes between
benign social bots, which were designed with good intentions, and malicious bots. The aim behind
malicious social bots may include altering other humans’ behaviour or distorting the opinion climate
in social media communities (Ferrara et al. 2014; Ross et al. 2019). However, examples in the recent
past have shown that, without proper monitoring, also benign bots can turn malicious. For instance, in
2016, Microsoft’s Tay had to be shut down due to obscene and inflammatory tweets it had learned
from other Twitter users (Neff and Nagy 2016). Furthermore, the unregulated nature of social media
entails the dissemination of information of questionable credibility (Wattal et al. 2010). The spread of
misinformation, defined as false or inaccurate information, especially that which is deliberately
intended to deceive (Lazer et al. 2018), poses a serious threat to public health during pandemics
(Zarocostas 2020). The amplification of the spreading of misinformation by social bots becomes
problematic if the health information and contributions, for example, contain information that can
influence the public opinion in a wrong way and thereby impede public health measures.
The distribution of health information, which is not consistent with information from evidence-based
sources, is a problem that emerged at the beginning of this century (Eysenbach 2002; Kim and Dennis
2019). Since then, the quantity and dissemination of misinformation have grown in unprecedented
ways (Shahi et al. 2020). Information epidemiology, or infodemiology, is an emerging research field,
which focuses on the determinants and distribution of health information in an electronic medium
and specifically the internet. It aims to identify areas of knowledge translation gaps between the best
evidence provided by experts and what most people believe (Eysenbach 2002). Research in this area
remains crucial since the diffused health information can alter the effectiveness of the
countermeasures taken by the government and also spread hysteria which can have a negative impact
on the mental well-being of social media users (Rosenberg et al. 2020). Within the information
systems discipline, a debate has been launched that questions the ability of IS research to actually help
fight pandemics such as COVID-19. Scholarship that contributes to infodemiology and provides
knowledge about ‘infodemics’ containment, we argue, can make an important contribution. Thus, in
this study, we focus on the detection of social bots with a malicious intent, i.e. spreading rumors and
unverified information. The research is guided by the following research question:
RQ: To what extend do social bots affect the Twitter discussion about COVID-19 by disseminating
In order to answer this research question, we apply social bot detection techniques based on three
metrics: Tweet Uniqueness (TU), Tweet Frequency (TFQ), and Friends-Followers-Ratio (FFR). The
underlying data set covers a period of 12 weeks of twitter communication about the COVID-19
pandemic and includes 3,052,030 tweets authored by 542,345 users. Subsequently to the bot
detection, a manual content analysis was conducted to identify misinformation among the social bot
content. Consequently, this study aims to contribute to scholarship on infodemiology by characterizing
social bots and their role during an ‘infodemic’. We further provide a basis of discussion about social
bot detection and provide implications for practitioners to better control malicious social bot content.
In the remainder, we provide a literature background, explain our methodical approach and present
preliminary results. To conclude this paper, the next steps of this research-in-progress are presented.
2 Related Work
2.1 Infodemiology
The term infodemiology is a portmanteau of information and epidemiology and can be defined as the
science of distribution and determinants of disease in populations and information in an electronic
medium, specifically the Internet and social media (Eysenbach, 2009). The latter is defined asa
group of Internet-based applications that build on the ideological and technological foundations of
Web 2.0, and that allow the creation and exchange of User Generated Content. (Kaplan and
Haenlein 2010). It has the ultimate aim to provide public health professionals, researchers, and policy
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
makers with the required tools and information to influence public health and policy decisions
(Eysenbach 2002, 2009). This research area initially focused on the identification of misinformation.
Since then, the term has been used to analyse the relationship between the demand for health
information, which can be investigated for example through web queries analysis, and the supply of
health information, for which supply-based applications are carried out like social media analysis
(Zeraatkar and Ahmadi 2018).
Metrics and approaches of infodemiology are an integral part of health informatics, the most popular
sources being Twitter and Google. The potential and feasibility of using social media to conduct studies
with an infodemiological approach has been demonstrated in previous studies (Mavragani 2020;
Mavragani et al. 2018). Eysenbach (2002), who first proposed the term ‘infodemiology’, suggested
during the SARS pandemic that the use of population health technologies, such as the Internet, can
help detect outbreaks of the disease at an early stage. During the H1N1 pandemic, an infodemiological
approach was used called ‘infoveillance’, which describes the usage of infodemiology methods for
surveillance purposes. This study illustrated the potential of using social media to conduct
infodemiological studies and showed that H1N1-related tweets on Twitter were primarily used for the
dissemination of information from credible sources, but also included many opinions and experiences
(Chew and Eysenbach 2010). Another infoveillance study on this subject matter identified main topics
that were most discussed by Twitter users and additionally analysed the sentiments of the tweets (Abd-
Alrazaq et al. 2020).
2.2 Social Bots
Typically, crises communication on social media is dominated by customary actors such as individuals
from the general public, affected organisations, and influential individuals such as politicians,
journalists, and influencers (Mirbabaie and Zapatka 2017; Stieglitz, Bunker, et al. 2017). Emerging
actors in this context are social bots (Brachten et al. 2018). The identification of bot accounts is of
increasing interest in research and society, especially on Twitter (Bruns et al. 2018). Recent studies
have been conducted to describe how bots are used to amplify messages on social media (Brünker et al.
2020; Vosoughi et al. 2018) or how fake content is populated on Twitter (Gupta et al. 2013). In
general, social bots are automated accounts that exhibit human-like behaviour with either malicious or
benign objectives (Ferrara et al. 2014). The detection of social bots has become more complex because
social bots become increasingly sophisticated. In this respect, it has become more difficult for
individuals to distinguish between a human and a social bot account (Freitas et al. 2015). Considering
that between 9% and 15% of active Twitter accounts might be bots (Varol et al. 2017), the detection of
social bots on social media poses a fundamental challenge.
Recent work has also focused on how social bots can bias discussions within social networks (Ferrara
et al. 2014; Ross et al. 2019). For example, social bots successfully influenced political discussions and
even had an effect on the outcomes of elections (Brachten et al. 2017; Ferrara 2017). Social bots are
also applied for spamming and thereby diverting attention from discussed issues (Bradshaw and
Howard 2017) or overstating trends (Brachten et al. 2018). For this purpose, social bot strategies like
smoke screening, astroturfing, and misdirecting are used (Stieglitz, Brachten, et al. 2017). Messias et
al. (2013) and Zhang et al. (2013) show that social bots can cooperate to manipulate the influence
scores of several centrality measures. Twitter, for instance, already took countermeasures against bots
with the goal of reducing the impact of such accounts’ malicious actions and therefore targeted
accounts that had been actively used to amplify and disseminate news of questionable sources
(Onuchowska et al. 2019). However, with unprecedented amounts of communication and emerging
phenomena such as the COVID-19 ‘infodemic’, it remains imperative to conduct research that aims at
understanding the technological enablers and misinformation mechanisms behind social bot activity.
3 Research Design
3.1 Case Description
After the first appearance of the infectious respiratory coronavirus disease in Wuhan, China, it vastly
spread in China and other countries around the world (World Health Organization 2020a). The
disease is also known as COVID-19 and causer of a global crisis. For Germany, the virus has not been
as unpredictable as other emerging infectious diseases because in Germany the first cases have been
confirmed on January 28, 2020 while in China the first cases have been confirmed on January 11
(World Health Organization 2020b). The opportunity for individuals to share their own stories,
feelings, opinions, judgments, or evaluations about the virus on social media channels did not just lead
to an increase of individual activity but also social bot activity on online social platforms such as
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
Twitter, Facebook, and YouTube (Abd-Alrazaq et al. 2020; Laato et al. 2020). To further investigate
the spread of misinformation, an infodemiological approach is going to be conducted, where the idea
of the field is to measure the pulse of public opinion, attention, behaviour, knowledge and attitudes by
tracking what people do and write on the Internet (Eysenbach 2009).
3.2 Data Collection
The data used in this study was collected via the Twitter API which is a commonly used tracking
method since it enables researchers to retrieve a wide range of meta-data (Stieglitz, Mirbabaie, et al.,
2018). Overall, data was crawled during the time period of February 27 (0:00 UTC) to May 20, 2020
(23:59 UTC) by applying a self-developed Java crawler using the Twitter4J1 library and saving it in a
MySQL database. In general, three separate crawling processes were conducted because new relevant
hashtags and terms appeared in the public discourse over the course of time, for example, after the
coronavirus got an official name by the WHO in February. The first crawling process tracked the
following terms: Coronavirus, nCoV2019, WuhanCoronaVirus, WuhanVirus, CoronaVirusOutbreak,
Ncov2020 and coronaviruschina. To retrieve a reliable dataset, a second and third data crawling were
conducted by tracking tweets using at least one of the terms Covid19, covid-19, covid_19, sarscov2,
covid, cov, corona, corona Germany, #COVID19de, #coronapocalypse, #staythefuckhome,
#flattenthecurve, and #stopthespread. The third data crawling contained specific terms which were
especially popular in German-speaking tweets. Finally, all three crawled data sets were merged, and
duplicates were removed, representing the foundation for this research. The tracking yielded a total of
542,345 users creating 3,052,030 tweets. A directed retweet network for further analyses is created,
consisting of users (vertices) and retweets (edges).
3.3 Social Bot Detection
Only highly active users are considered for the creation of a sample in this research because the activity
is also reflected in the likelihood of the account to be a bot (Ferrara et al. 2014). To this end, accounts
which posted at least 150 tweets within two weeks during the tracking period were considered highly
active. Tweets, retweets and mentions are included in the dataset to cover the entire spectrum of
Twitter communication. For the identification of bot-like accounts, the dataset was preprocessed, i.e.
deletion of unnecessary meta data and exclusion of inactive users. The social bot identification as well
as the preprocessing were performed by using R which is a free software environment for statistical
computing and graphics (Nordhausen 2015). Three metrics are considered for the identification of the
most social bot-like users: (1) the Tweet Uniqueness (TU), (2) the Tweet Frequency (TFQ), and (3) the
Friends-Follower-Ratio (FFR) and for every metric, a specific threshold was derived to create a sample
of potential social bots (Brünker et al., 2020).
Figure 1. Social Bot Sampling Procedure (adapted by Brünker et al., 2020).
The TU quantifies the particularity of the content of a tweet. For the computation of this measure, the
number of distinct tweets is divided by the total number of tweets k by the user, according to (Ross et
al. 2018). A tweet is unique if the tweet alters in at least one character from every other user's tweet.
The TU will equal 1 if every tweet by a user is different from others and the more identical tweets a user
publishes the more this value decreases. The global TU in the dataset, which is the number of unique
tweets divided by the total number of extracted tweets, serves as the threshold every user must fall
below in order to be classified as a potential bot (82,692/117,678 = 0.703).
A high tweet frequency is and indicator for bot-like behaviour (Brachten et al. 2017; Varol et al. 2017).
The TFQ displays the activity of a user during the tracking period and is calculated by dividing the total
number of tweets of the user by the number of hours of the tracking period. For example, a TFQ of 3
means that the user publishes three tweets per hour each day. In this regard, the threshold of 1.042
must be exceeded in order for the user to be assigned to the social bot sample. In this case it means
that a user must have disseminated at least 25 tweets a day during the tracking period.
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
The FFR indicates the relation between the friends and the followers of a user. In recent studies, the
relation between a users’ friends and followers was considered for social bot detection (Chu et al. 2012;
Ferrara et al. 2014). The relationships on Twitter are unidirectional, i.e. if a user A follows the user B,
then A is a follower of B and A’s friend is B. The ratio is calculated by dividing the total number of
friends by the total number followers of a user and the result must exceed the threshold of a ratio of
3.5. Calculating a global FFR resulted in a way too low value which has led to a congested social bot
sample because most of the active accounts have been assigned to the sample. The minimum FFR was
set 3.5 so that, for example accounts with 1,000 followers must be friends with at least 3,500 other
users in order to be classified as potential bots.
4 Preliminary Results
Overall, 380 accounts, who are responsible for 117,678 tweets, are considered highly active within the
dataset. 20,5% (78 accounts) of them meet at least one of the conditions and are therefore classified as
likely social bots. These, in turn, disseminated 19,117 tweets throughout a 12-week time period. The
manual content analysis has been conducted for the 78 classified accounts.
Table 1 shows selected examples of tweets by some of the alleged bots. The tweets contain
misinformation or conspiracy content, but the mentioned users did not disseminate this kind of
content solely. In particular, there are also retweets of news and updates about the virus. None of the
accounts in the social bot sample could be classified as a social bot based on a single tweet. However,
analysing the account’s characteristics according the metrics TU, TFQ or FFR allows to better classify
the distinct accounts. Particularly some accounts significantly surpassing the threshold of the metric
TFQ. For instance, one account created 3,967 tweets, including news, many hashtags, and links to
related websites or retweeted his own tweets.
RT @WaltiSiegrist: The worldwide biological weapons research of the
USA #SARSCoV2 #coronavirus #covid19
RT @lobbycontrol: Did #party donations play a role in the occupation
of the #NRW-#Corona Council? More than 1.1 million Euro flowed
2002-2017 from the company #Trumpf and its owners to #CDU
(765.766 €) and #FDP (335.000 €). Trumpf is represented on the
council by the billionaire Nicola #Leibinger-Kammüller
RT @hfeldwisch: How #Taiwan prevented the #COVID19 outbreak in
his country - and the @WHO doesn't want to know about it: Exciting
article by Richard Friebe @TspWissenschaft
Table 1. Selected examples of social bots spreading misinformation or conspiracy theories.
5 Conclusion and Next Steps
The aim of this research is to determine the extent of the impact social bots have on the network by
disseminating misinformation or conspiracy theories and to provide an evaluation on whether the
identified social bots can be considered a threat measured by their influence. Particularly, in this
global pandemic, social bots have been neglected so far and the objective of this research is to gain new
insights into how social bots are applied in this context. For this purpose, this research-in-progess
carried out the first steps of data collection, preprocessing and manual analyzing.
The manual assessment of bot-like accounts that entered the sample as they passed over the
thresholds defined by Brünker et al. (2020) revealed that their impact on the COVID-19 ‘infodemic’ is
very limited. However, the analysis provides valuable insights that help us to better understand online
behaviour that is relevant to an ‘infodemics’ context. Those mechanisms do not only apply to social
bots but can be transferred onto other social media users trying to exert influence within an
‘infodemic’ environment.
However, the number of relevant tweets grew during the data acquisition process. As a result, some of
the tweets among the hashtags examined were not tracked in time and no absolute completeness of the
data set can be guaranteed. To properly analyse the high number of tweets in our dataset, an
automated content analysis may be applied as part of the future work. Since “automated accounts are
particularly active in the early spreading phases of viral claims” (Shao et al., 2017, p. 11) and there is
missing data between the first infection and the beginning of the data crawling, namely January 11 to
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
February 27, it is possible that many social bots were not caught or already blocked or deleted by
Twitter itself.
To analyse the impact of these social bot accounts on the dissemination of misinformation and
conspiracy theories, their individual PageRank value within a retweet network can be calculated and
compared. As a next step, the PageRank score can be evaluated within a created retweet network of the
identified accounts. PageRank is a variation of the Eigenvector centrality measure which considers
network information, i.e. users who are closely related to influential users, to quantify the influence of
a node (Bonacich 1972). It is developed for directed graphs such as retweet networks and integrates
the flow of influence to calculate the authority of a node in a network (Yang and Tang 2012). The
PageRank scores of the social bots can be compared to other users’ scores and on the basis of this,
statements about the impact of the social bots throughout the network can be derived.
Moreover, the tweets of social bots are likely to be retweeted because misinformation spreads faster
than information from credible sources (Shu et al. 2020). Also, social bots can cooperate and thereby
reach a higher PageRank value than other users (Messias et al. 2013; Zhang et al. 2013). In this case
and additionally, a tweet of a social bot is retweeted by other influential users, the bot is very likely to
be considered influential as well. Depending on the results of the PageRank algorithm and the
identified bots being influential or not, they can be classified as threat or harmless.
6 References
Abd-Alrazaq, A., Alhuwail, D., Househ, M., Hai, M., and Shah, Z. 2020. “Top Concerns of Tweeters
during the COVID-19 Pandemic: A Surveillance Study,” Journal of Medical Internet Research
(22:4), pp. 19. (
Bonacich, P. 1972. “Factoring and Weighting Approaches to Status Scores and Clique Identification,”
The Journal of Mathematical Sociology (2:1), pp. 113120.
Brachten, F., Mirbabaie, M., Stieglitz, S., Berger, O., Bludau, S., and Schrickel, K. 2018. “Threat or
Opportunity? - Examining Social Bots in Social Media Crisis Communication,” in Australasian
Conference on Information Systems, Sydney.
Brachten, F., Stieglitz, S., Hofeditz, L., Kloppenborg, K., and Reimann, A. 2017. “Strategies and
Influence of Social Bots in a 2017 German State Election - A Case Study on Twitter,” in
Australasian Conference on Information Systems.
Bradshaw, S., and Howard, P. 2017. “Troops, Trolls and Troublemakers: A Global Inventory of
Organized Social Media Manipulation,” The Lancet (274:7095), p. 187.
Brünker, F., Marx, J., Mirbabaie, M., and Stieglitz, S. 2020. “‘ The Tireless Selling-Machine ’
Commercial Deployment of Social Bots during Black Friday Season on Twitter,” in Proceedings
of the 15th International Conference on Wirtschaftsinformatik, Potsdam, Germany.
Bruns, A., Moon, B., Münch, F. V., Wikström, P., Stieglitz, S., Brachten, F., and Ross, B. 2018.
“Detecting Twitter Bots That Share SoundCloud Tracks,” ACM International Conference
Proceeding Series, pp. 251255. (
Chew, C., and Eysenbach, G. 2010. “Pandemics in the Age of Twitter: Content Analysis of Tweets
during the 2009 H1N1 Outbreak,” PLoS ONE (5:11).
Chu, Z., Gianvecchio, S., Wang, H., and Jajodia, S. 2012. “Detecting Automation of Twitter Accounts:
Are You a Human, Bot, or Cyborg?,” IEEE Transactions on Dependable and Secure Computing
(9:6), pp. 811824. (
Eysenbach, G. 2002. Infodemiology: The Epidemiology of (Mis)Information, (9343:02), pp. 763765.
Eysenbach, G. 2009. “Infodemiology and Infoveillance: Framework for an Emerging Set of Public
Health Informatics Methods to Analyze Search, Communication and Publication Behavior on
the Internet.,” Journal of Medical Internet Research (11:1), pp. 110.
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
Ferrara, E. 2017. “Disinformation and Social Bot Operations in the Run up to the 2017 French
Presidential Election,” First Monday (22:8). (
Ferrara, E., Varol, O., Davis, C., Menczer, F., and Flammini, A. 2014. “The Rise of Social Bots,”
Communication of the ACM (59:7), pp. 96104. (
Freitas, C., Benevenuto, F., Ghosh, S., and Veloso, A. 2015. “Reverse Engineering Socialbot Infiltration
Strategies in Twitter,” Proceedings of the 2015 IEEE/ACM International Conference on
Advances in Social Networks Analysis and Mining, ASONAM 2015, pp. 2532.
Gupta, A., Lamba, H., and Kumaraguru, P. 2013. “$1.00 per RT #BostonMarathon #PrayForBoston:
Analyzing Fake Content on Twitter,” ECrime Researchers Summit, ECrime.
Kaplan, A. M., and Haenlein, M. 2010. “Users of the World, Unite! The Challenges and Opportunities
of Social Media,Business Horizons (53:1), p. 61.
Kim, A., and Dennis, A. 2019. “Says Who? The Effects of Presentation Format and Source Rating on
Fake News in Social Media,” MIS Quarterly (43), pp. 10251039.
Laato, S., Islam, A. K. M. N., Islam, M. N., and Whelan, E. 2020. “What Drives Unverified Information
Sharing and Cyberchondria during the COVID-19 Pandemic?,” European Journal of
Information Systems (29:3), Taylor & Francis, pp. 288305.
Lazer, D. M. J., Baum, M. A., Benkler, Y., Berinsky, A. J., Greenhill, K. M., Menczer, F., Metzger, M. J.,
Nyhan, B., Pennycook, G., Rothschild, D., Schudson, M., Sloman, S. A., Sunstein, C. R., Thorson,
E. A., Watts, D. J., and Zittrain, J. L. 2018. “The Science of Fake News,” Science (359:6380), pp.
10941096. (
Mavragani, A. 2020. “Tracking COVID-19 in Europe: Infodemiology Approach,” JMIR Public Health
and Surveillance (6:2), p. e18941. (
Mavragani, A., Ochoa, G., and Tsagarakis, K. P. 2018. “Assessing the Methods, Tools, and Statistical
Approaches in Google Trends Research: Systematic Review,” Journal of Medical Internet
Research (20:11), pp. 120. (
Messias, J., Schmidt, L., Oliveira, R. A. R., and Benevenuto, F. 2013. “You Followed My Bot!
Transforming Robots into Influential Users in Twitter,” First Monday (18:7), pp. 015.
Mirbabaie, M., Bunker, D., Stieglitz, S., and Deubel, A. 2020. “Who Sets the Tone? Determining the
Impact of Convergence Behaviour Archetypes in Social Media Crisis Communication,”
Information Systems Frontiers (22:2), Information Systems Frontiers, pp. 339351.
Mirbabaie, M., and Marx, J. 2019. “‘Breaking’ News: Uncovering Sense-Breaking Patterns in Social
Media Crisis Communication during the 2017 Manchester Bombing,” Behaviour and
Information Technology (39:3), Taylor & Francis, pp. 252266.
Mirbabaie, M., and Zapatka, E. 2017. “Sensemaking in Social Media Crisis Communication - A Case
Study on the Brussels Bombings in 2016,” in Proceedings of the 25th European Conference on
Information Systems, pp. 21692186.
Neff, G., and Nagy, P. 2016. “Talking to Bots: Symbiotic Agency and the Case of Tay,” International
Journal of Communication (10:1), pp. 49154931.
Nordhausen, K. 2015. “Statistical Analysis of Network Data with R,” International Statistical Review,
pp. 160174. (
Onuchowska, A., Berndt, D. J., and Samtani, S. 2019. “Rocket Ship or Blimp? – Implications of
Malicious Accounts Removal on Twitter,” in Proceedings of the 27th European Conference on
Information Systems, pp. 1-10.
Australasian Conference on Information Systems Marx et al
2020, Wellington Social Bots and Infodemics
Rosenberg, H., Syed, S., and Rezaie, S. 2020. “The Twitter Pandemic: The Critical Role of Twitter in
the Dissemination of Medical Information and Misinformation during the COVID-19
Pandemic,” Cjem (May), pp. 17. (
Ross, B., Brachten, F., Stieglitz, S., Wikström, P., Moon, B., Münch, F. V., and Bruns, A. 2018. “Social
Bots in a Commercial Context A Case Study on Soundcloud,” in Twenty-Sixth European
Conference on Information Systems (ECIS2018), Portsmouth, UK.
Ross, B., Pilz, L., Cabrera, B., Brachten, F., Neubaum, G., and Stieglitz, S. 2019. “Are Social Bots a Real
Threat? An Agent-Based Model of the Spiral of Silence to Analyse the Impact of Manipulative
Actors in Social Networks,” European Journal of Information Systems, Taylor & Francis.
Shahi, G. K., Dirkson, A., and Majchrzak, T. A. 2020. “An Exploratory Study of COVID-19
Misinformation on Twitter.” (
Shao, C., Ciampaglia, G. L., Varol, O., Yang, K. C., Flammini, A., and Menczer, F. 2017. “The Spread of
Fake News by Social Bots,” Nature Communications (9:1). (
Shu, K., Mahudeswaran, D., Wang, S., Lee, D., and Liu, H. 2020. “FakeNewsNet: A Data Repository
with News Content, Social Context, and Spatiotemporal Information for Studying Fake News on
Social Media,” Big Data (8:3), pp. 171188. (
Stieglitz, S., Brachten, F., Ross, B., and Jung, A.-K. 2017. Do Social Bots Dream of Electric Sheep? A
Categorisation of Social Media Bot Accounts, pp. 111.
Stieglitz, S., Bunker, D., Mirbabaie, M., and Ehnis, C. 2017. “Sense-Making in Social Media during
Extreme Events,” Journal of Contingencies and Crisis Management (25:3), pp. 112.
Varol, O., Ferrara, E., Davis, C. A., Menczer, F., and Flammini, A. 2017. “Online Human-Bot
Interactions: Detection, Estimation, and Characterization,” in Proceedings of the Eleventh
International AAAI Conference on Web and Social Media (ICWSM 2017).
Vosoughi, S., Roy, D., and Aral, S. 2018. “The Spread of True and False News Online,” Science (1151:3),
pp. 11461151.
Wattal, S., Schuff, D., Mandviwalla, M., and Williams, C. B. 2010. “Web 2.0 and Politics: The 2008
U.S. Presidential Election and an e-Politics Research Agenda,” MIS Quarterly: Management
Information Systems (34:4), pp. 669688. (
World Health Organization. 2020a. “Novel Coronavirus (2019-NCoV) Situation Report - 1,” World
Health Organization Bulletin, pp. 17.
World Health Organization. 2020b. “WHO Coronavirus Disease (COVID-19) Dashboard.”
Yang, C. C., and Tang, X. 2012. “Estimating User Influence in the MedHelp Social Network,” IEEE
Intelligent Systems (27:5), pp. 4450. (
Zarocostas, J. 2020. “How to Fight an Infodemic,” Lancet (London, England) (395:10225), Elsevier
Ltd, p. 676. (
Zeraatkar, K., and Ahmadi, M. 2018. “Trends of Infodemiology Studies: A Scoping Review,” Health
Information and Libraries Journal (35:2), pp. 91120. (
Zhang, J., Zhang, R., Zhang, Y., and Yan, G. 2013. “On the Impact of Social Botnets for Spam
Distribution and Digital-Influence Manipulation,” 2013 IEEE Conference on Communications
and Network Security, CNS 2013, pp. 4654. (
This project has received funding from the European Union’s Horizon 2020 research and innovation
programme under the Marie Skłodowska-Curie grant agreement No 823866.
Copyright © 2020 Julian Marx, Felix Brünker, Milad Mirbabaie, Eric Hochstrate. This is an open-
access article licensed under a Creative Commons Attribution-NonCommercial 3.0 New Zealand,
which permits non-commercial use, distribution, and reproduction in any medium, provided the
original author and ACIS are credited.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
During the COVID-19 pandemic, social media has become a home ground for misinformation. To tackle this infodemic, scientific oversight, as well as a better understanding by practitioners in crisis management, is needed. We have conducted an exploratory study into the propagation, authors and content of misinformation on Twitter around the topic of COVID-19 in order to gain early insights. We have collected all tweets mentioned in the verdicts of fact-checked claims related to COVID-19 by over 92 professional fact-checking organisations between January and mid-July 2020 and share this corpus with the community. This resulted in 1500 tweets relating to 1274 false and 226 partially false claims, respectively. Exploratory analysis of author accounts revealed that the verified twitter handle(including Organisation/celebrity) are also involved in either creating(new tweets) or spreading(retweet) the misinformation. Additionally, we found that false claims propagate faster than partially false claims. Compare to a background corpus of COVID-19 tweets, tweets with misinformation are more often concerned with discrediting other information on social media. Authors use less tentative language and appear to be more driven by concerns of potential harm to others. Our results enable us to suggest gaps in the current scientific coverage of the topic as well as propose actions for authorities and social media users to counter misinformation.
Full-text available
The World Health Organisation has emphasised that misinformation – spreading rapidly through social media – poses a serious threat to the COVID-19 response. Drawing from theories of health perception and cognitive load, we develop and test a research model hypothesising why people share unverified COVID-19 information through social media. Our findings suggest a person’s trust in online information and perceived information overload are strong predictors of unverified information sharing. Furthermore, these factors, along with a person’s perceived COVID-19 severity and vulnerability influence cyberchondria. Females were significantly more likely to suffer from cyberchondria, with males more likely to share news without verifying its reliability. Our findings suggest that to mitigate the spread of COVID-19 misinformation and cyberchondria, measures should be taken to enhance a healthy scepticism of health news while simultaneously guarding against information overload.
Full-text available
Social media has become a popular means for people to consume and share the news. At the same time, however, it has also enabled the wide dissemination of fake news, that is, news with intentionally false information, causing significant negative effects on society. To mitigate this problem, the research of fake news detection has recently received a lot of attention. Despite several existing computational solutions on the detection of fake news, the lack of comprehensive and community-driven fake news data sets has become one of major roadblocks. Not only existing data sets are scarce, they do not contain a myriad of features often required in the study such as news content, social context, and spatiotemporal information. Therefore, in this article, to facilitate fake news-related research, we present a fake news data repository FakeNewsNet, which contains two comprehensive data sets with diverse features in news content, social context, and spatiotemporal information. We present a comprehensive description of the FakeNewsNet, demonstrate an exploratory analysis of two data sets from different perspectives, and discuss the benefits of the FakeNewsNet for potential applications on fake news study on social media.
Full-text available
Background: Infodemiology, i.e. information epidemiology, uses Web-based data in order to inform public health and policy. Infodemiology metrics have been widely and successfully employed in order to assess and forecast epidemics and outbreaks. Objective: In light of the recent COVID-19 pandemic that started in Wuhan, China, in 2019, in this report online search traffic data from Google are used aiming at tracking the spread of the new Coronavirus. Methods: Time-series from Google Trends from January to March 2020 on the topic of "Coronavirus" are retrieved and correlated with official data on COVID-19 cases and deaths in the European countries that have been affected the most; Italy (at national and regional level), Spain, France, Germany, and the UK. Results: Statistically significant correlations are observed between the online interest and COVID-19 cases and deaths. Furthermore, a critical point after which the Pearson correlation coefficient starts declining (even if it is still is statistically significant) is identified, indicating that this method is most efficient in regions or countries that have not peaked in COVID-19 cases yet. Conclusions: In the past, infodemiology metrics in general and data from Google Trends in specific, have been shown to be useful in tracking and forecasting outbreaks, epidemics and pandemics, as, for example, in the cases of MERS, Ebola, measles, and Zika. With the COVID-19 pandemic still at the beginning, it is essential to explore and combine new methods of disease surveillance, in order to assist with the preparedness of the respective health care systems at regional level.
Full-text available
The twitter pandemic: The critical role of twitter in the dissemination of medical information and misinformation during the COVID-19 Pandemic - Hans Rosenberg, Shahbaz Syed, Salim Rezaie
Full-text available
Individuals, (media-) organisations, and crisis responders who are involved in ad hoc crisis communication steadily deploy social media to contribute to collective sense-making as an endeavour to create meaning in highly uncertain situations. Exerting sense-giving in order to shape others’ conceptions is causally preceded by an initial breakup of existing understanding. This study aims to explore patterns of sense-breaking in social media crisis communication and its impact on collective sense-making and sense-giving. To this end, we conducted a case study of the Manchester bombing in 2017, including a social network analysis of 708,147 Twitter postings and a content analysis of 2,006 original tweets. We found individual role types to be initiators of sense-breaking in early crisis stages when uncertainty is at its height. Exerting successive sense-giving becomes more challenging if the collective sense-making has progressed along with the sequence of events. This understanding aims to encourage emergency management organisations to move their sense-giving actions closer to the point in time when sense-breaking occurs.
Full-text available
Information systems such as social media strongly influence public opinion formation. Additionally, communication on the internet is shaped by individuals and organisations with various aims. This environment has given rise to phenomena such as manipulated content, fake news, and social bots. To examine the influence of manipulated opinions, we draw on the spiral of silence theory and complex adaptive systems. We translate empirical evidence of individual behaviour into an agent-based model and show that the model results in the emergence of a consensus on the collective level. In contrast to most previous approaches, this model explicitly represents interactions as a network. The most central actor in the network determines the final consensus 60–70% of the time. We then use the model to examine the influence of manipulative actors such as social bots on public opinion formation. The results indicate that, in a highly polarised setting, depending on their network position and the overall network density, bot participation by as little as 2–4% of a communication network can be sufficient to tip over the opinion climate in two out of three cases. These findings demonstrate a mechanism by which bots could shape the norms adopted by social media users.
Full-text available
Background: In the era of information overload, are big data analytics the answer to access and better manage available knowledge? Over the last decade, the use of Web-based data in public health issues, that is, infodemiology, has been proven useful in assessing various aspects of human behavior. Google Trends is the most popular tool to gather such information, and it has been used in several topics up to this point, with health and medicine being the most focused subject. Web-based behavior is monitored and analyzed in order to examine actual human behavior so as to predict, better assess, and even prevent health-related issues that constantly arise in everyday life. Objective: This systematic review aimed at reporting and further presenting and analyzing the methods, tools, and statistical approaches for Google Trends (infodemiology) studies in health-related topics from 2006 to 2016 to provide an overview of the usefulness of said tool and be a point of reference for future research on the subject. Methods: Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines for selecting studies, we searched for the term “Google Trends” in the Scopus and PubMed databases from 2006 to 2016, applying specific criteria for types of publications and topics. A total of 109 published papers were extracted, excluding duplicates and those that did not fall inside the topics of health and medicine or the selected article types. We then further categorized the published papers according to their methodological approach, namely, visualization, seasonality, correlations, forecasting, and modeling. Results: All the examined papers comprised, by definition, time series analysis, and all but two included data visualization. A total of 23.1% (24/104) studies used Google Trends data for examining seasonality, while 39.4% (41/104) and 32.7% (34/104) of the studies used correlations and modeling, respectively. Only 8.7% (9/104) of the studies used Google Trends data for predictions and forecasting in health-related topics; therefore, it is evident that a gap exists in forecasting using Google Trends data. Conclusions: The monitoring of online queries can provide insight into human behavior, as this field is significantly and continuously growing and will be proven more than valuable in the future for assessing behavioral changes and providing ground for research using data that could not have been accessed otherwise.
Conference Paper
Full-text available
Recently, automated communication on social media has seen increased attention. Social bots, social media accounts controlled by algorithms that mimic human behaviour, have been found to attempt to influence users in several political contexts. However, their use in a commercial context, e.g. to boost sales of a product by aggressively promoting it with thousands of messages, has so far been neglected. To address this shortcoming, this paper examines the case of the social media music platform SoundCloud. We gathered a dataset of six months of activity, comprising 15,850,069 tracks and 12,125,095 comments. We then calculated a comment uniqueness score for highly active accounts to assess the variability of their comments. First analyses show that some accounts post suspiciously repetitive comments. These accounts also frequently repost existing content, but contribute little original content. An analysis of the commenting network further underlines that these accounts differ clearly from regular users. We conclude that the comment uniqueness metric can be used as an indicator to distinguish bots from humans, and that a considerable proportion of SoundCloud comments are likely to emanate from bots or semi-automated accounts. The implications of these findings and future plans are discussed.