Digital Lighthouse: A Platform for Monitoring Public Groups
in WhatsApp
Ivandro Claudino de S
´
a
1
, Jos
´
e Maria Monteiro
1
, Jos
´
e Wellington Franco da Silva
1
,
Leonardo Monteiro Medeiros
1
, Pedro Jorge Chaves Mour
˜
ao
2
and Lucas Cabral Carneiro da Cunha
1
1
Computer Science Department, Federal University of Cear
´
a, Fortaleza, Cear
´
a, Brazil
2
Department of Sociology, Cear
´
a State University, Fortaleza, Cear
´
a, Brazil
pjmourao cs@hotmail.com, lucascabral@aridalab.dc.ufc.br
Keywords:
Misinformation Detection, Natural Language Processing, WhatsApp, Social Media.
Abstract:
The large-scale dissemination of misinformation through social media has become a critical issue, harming
social stability, democracy, and public health. In Brazil, 48% of the population uses WhatsApp to get news.
So, many groups have been used this instant messaging application to spread misinformation, especially as
part of articulated political or ideological campaigns. In this context, WhatsApp provides an important feature:
the public groups. These groups are so suitable for misinformation dissemination. Thus, developing software
frameworks to monitor the misinformation spreading in WhatsApp public groups has become a field of high
interest both in academia, government and industry. In this work, we present an entire platform, called Dig-
ital Lighthouse, that aims for finding WhatsApp public groups, besides extracting, cleaning, analyzing, and
visualizing misinformation that spread in such groups. Using the Digital Lighthouse, we built three different
datasets. We hope that our platform can help journalists and researchers to understand the misinformation
propagation in Brazil.
1 INTRODUCTION
In the last years, the popularity of instant messaging
applications has contributed to the spread of misinfor-
mation. Through these systems, misinformation can
deceive thousands of people in a short time (due to
their appealing nature) and cause significant harm to
individuals or society. In this context, misinformation
has been used to change political scenarios, to con-
tribute to the spread of diseases, and even to cause
deaths (Su et al., 2020).
The WhatsApp instant messaging application is
very popular in Brazil, with more than 120 million
users. In Brazil, 48% of the population use WhatsApp
to get, share and discuss news. WhatsApp makes
it possible to instantly share different media types,
such as images, audios, and videos. Besides, What-
sApp provides a significant feature: the public groups.
These public groups are accessible through invitation
links published on popular websites and various so-
cial networks, such as Facebook and Twitter. Usually,
they have specific topics for discussion, such as pol-
itics and education. In this way, WhatsApp public
groups are very similar to social networks
Public groups have been used to spread misinfor-
mation, especially as part of articulated political or
ideological campaigns. Furthermore, misinformation
spreads faster, deeper, and expansive than legit infor-
mation. Further, due to the high volume of informa-
tion that we are exposed to, we have a limited ability
to distinguish true information from misinformation
(Vosoughi et al., 2018; Qiu et al., 2017).
In this context, monitoring the content that cir-
culates in public WhatsApp groups is a fundamental
task to understand the spread of misinformation and
get insights to address this problem. However, col-
lecting a database of WhatsApp messages is a chal-
lenging task. To fill this gap, we built the Digital
Lighthouse, an entire platform that aims for finding
WhatsApp public groups, besides extracting, clean-
ing, analyzing, and visualizing misinformation that
spread in these groups. Early detection of misin-
formation could prevent its spread, thus reducing its
damage. Using the Digital Lighthouse, we build three
different WhatsApp’ messages datasets, covering rel-
evant themes such as the Brazilian general elections
campaign in 2018, the covid-19 pandemic, and the
vaccine for covid-19.
Claudino de Sá, I., Monteiro, J., Franco da Silva, J., Medeiros, L., Mourão, P. and Carneiro da Cunha, L.
Digital Lighthouse: A Platform for Monitoring Public Groups in WhatsApp.
DOI: 10.5220/0010480102970304
In Proceedings of the 23rd International Conference on Enterprise Information Systems (ICEIS 2021) - Volume 1, pages 297-304
ISBN: 978-989-758-509-8
Copyright
c
2021 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
297
The remainder of this paper is organized as fol-
lows. Section 2 presents the main related work. Sec-
tion 3 describes the Digital Lighthouse platform. Sec-
tion 4 details a case study performed to evaluate the
proposed platform. Conclusions and future work are
presented in Section 5.
2 RELATED WORK
It is essential to highlight that WhatsApp is unique
in several ways relative to other social media plat-
forms. WhatsApp was developed to allow users to
privately send messages to each other through their
smartphones. A specific aspect of WhatsApp messag-
ing is the public groups. These are openly accessible
groups, frequently publicized on well-known web-
sites, and typically themed around particular topics.
It is worth mentioning that texts extracted from What-
sApp are quite different from those collected through
Websites, fact-checkers, or other kinds of social me-
dia platforms, such as Twitter. WhatsApp messages
include conversation, opinions, humorous and satir-
ical texts, prayers, commercial offers, news, short
texts, emojis, and others.
Thus, despite the scientific community’s efforts,
there is still a need for monitoring and identifying
misinformation in WhatsApp messages, mainly in
Portuguese. The paper presented in (Garimella and
Tyson, 2018) is a seminal work in collecting and
analyzing WhatsApp messages. The authors built
a dataset by crawling 178 public groups, containing
45K users and 454K messages, from different coun-
tries and languages, such as India, Pakistan, Russia,
Brazil, and Colombia. In (Gaglani et al., 2020), the
authors contextualize the problem of spreading fake
news on WhatsApp, especially in India and Brazil,
and proposes a strategy for the automatic detection of
fake news. A total of 10 public groups were scraped
for one week to get 1000 multilingual messages. In
(Resende et al., 2018), the authors presented a system
for gathering, analyzing, and visualize public groups
in WhatsApp. Besides, the authors also provide a
brief characterization of the 169.154 messages shared
by 6,314 users in 127 public groups. In the study pre-
sented in (Machado et al., 2019), the authors collected
and analyzed 298,892 WhatsApp messages, from 130
public groups, in the period of the 2018 Brazilian
presidential elections. In (Resende et al., 2019), the
authors analyzed different aspects of WhatsApp mes-
sages from public political-oriented groups. However,
none of these works provides an entire public plat-
form for finding, gathering, analyzing, and visualiz-
ing WhatsApp messages.
Other works propose classifiers to detect misin-
formation automatically (Silva et al., 2020; Faustini
and Cov
˜
oes, 2019). In (Shu et al., 2018), the authors
investigated the use of complex networks to detect
and mitigate fake news on social media. During fake
news dissemination, different entities can be catego-
rized into content, social and temporal dimensions.
These dimensions have mutual relations and depen-
dencies. So, fake news dissemination has inherent
network properties. In (Shu et al., 2019), the authors
explored user profiles to detect fake news. They argue
that there are correlations between malicious accounts
and fake news. In this same way, the paper presented
in (Hamdi et al., 2020) proposed a hybrid approach
that explores features from the user profile and his so-
cial graph (Twitter followers/followees graph) to de-
tect fake news. In (Zhang and Hara, 2020), the au-
thors propose a probabilistic model for malicious user
and rumor detection (MURD).
3 THE DIGITAL LIGHTHOUSE
PLATFORM
This section will present the main components of the
Digital Lighthouse platform, which aims to extract,
analyze, and visualize misinformation in WhatsApp
messages. The proposed platform architecture com-
prises four modules, as illustrated in Figure 1. The
main contribution of this work is the orchestration of
all these components, which will be detailed next.
3.1 Module I: Finding Public Groups
WhatsApp allows you to join public groups through
the use of links (URLs) containing the domain
’chat.whatsapp.com’ and a group identification code.
These links are publicized through websites or social
networks. In this way, groups can be found through
queries on search engines like Google, or simply by
accessing sites created for this specific purpose. This
work used both strategies for finding public groups.
3.1.1 Finding Web Pages with Invite Links
In order to find invitations links for WhatsApp public
groups through the Google search engine, we develop
a web crawler using the Python programming lan-
guage. The crawler builds queries, sends them to the
Google search engine and receives the result (links for
web pages). To set up a particular query, the crawler
receives a series of input parameters, sucha as: the
WhatsApp domain, a set of keywords, and the tar-
get language. After a given query be executed, the
ICEIS 2021 - 23rd International Conference on Enterprise Information Systems
298
Figure 1: The Digital Lighthouse Platform Architecture.
crawler receives a set of metadata, including refer-
ences to the web pages where the invite links were
found. These web page links are stored in a file called
search links.csv.
3.1.2 Finding Invite Links
The next step consists of requesting each web page
found previously (and stored in the search links.csv
file) and parse it seeking for WhatsApp invite links.
More specifically, the crawler sends a HTTP request
for a certain web page link (URL). Then, the search
engine answer the request by returning the web page
content. After, a scraper will create a tree structure
with the HTML content of the web page. This tree
structure will be used to search for invite links. Fi-
nally, the scraper produces as output a list of invite
links which are stored in a file called group links.csv
or yet list of non-filtered groups.
3.1.3 Selecting Valid Invite Links
However, find a set of invite links is not sufficient.
Some groups no longer exist, several links have been
disabled, and a few groups have a tiny number of par-
ticipants. Thus, it is necessary to check the status of
each invite link. After this checking process, a new
file, called a list of filtered groups, is generated con-
taining only the valid links.
3.1.4 Joining Public Groups
Finally, with valid links, it is possible to join public
groups using a cell phone chip and a web browser,
in an automatic or manual manner. In this work, we
manually joined the groups to don’t violate What-
sApp politics.
3.2 Module II: Getting and Storing
Data
Unlike other social media, such as Twitter and Face-
book, and due to its private chat nature, there is no
public API to collect data from WhatsApp in an au-
tomated manner. Thus, monitoring WhatsApp public
groups poses a technical and even ethical challenge.
To tackle this issue, we take an approach similar to
(Garimella and Tyson, 2018; Resende et al., 2018).
Thus, in order to automatically collect the content
(messages, audio, images, and videos) of the public
groups that Digital Lighthouse joined, it have used
WhatsApp Web and Selenium Web Driver.
3.2.1 Getting the Content of Public Groups
The Digital Lighthouse uses a virtual machine (VM)
containing an Android emulator, the WhatsApp Web,
the Selenium Web Driver and a PostgreSQL database
server. In the Android emulator we had installed the
Digital Lighthouse: A Platform for Monitoring Public Groups in WhatsApp
299
WhatsApp application and a SQLite database. Fi-
nally, we used the Selenium Web Driver to manipu-
late the Android emulator and the WhatsApp Web in
order to automatically ccess the public groups content
and store it in the SqlLite database.
3.2.2 Storing the Content of Public Groups
The messages extracted from WhatsApp are stored, in
their original format, in a SQLite database. However,
for that such messages can be effectively used for the
purpose of knowledge discovery or to get insights, it
is necessary that they undergo to a process of clean-
ing, integration and anonymization. After this pro-
cess, the treated messages are stored in a PostgreSQL
database, and can now be used for analysis and visu-
alization purposes. It is important to highlight that the
audios, images, and videos are stored in the file sys-
tem. The PostgreSQL database stores only the path to
these files.
A Python script was created to periodically per-
form the ETL process in order to clear, integrate,
anonymize and load messages from SQLite to Post-
greSQL database. We took into consideration pri-
vacy issues by anonymizing users’ names and cell
phone numbers. For this, we create an anonymous
and unique ID for each user by using an MD5 hash
function on its phone number. Similarly, we create
an anonymous alias for each group. Since the groups
are public, our approach does not violate WhatsApp’s
privacy policy
1
.
3.3 Module III: Knowledge Discovery
This module explores the data stored in the Post-
greSQL to finding implicit, previously unknown, and
potentially useful patterns. Its main component is the
Misinformation Detector, a machine learning classi-
fier once trained and tested. This component receives
a text as input and returns as output if the text is or
not the misinformation. Besides, two other compo-
nents are under development: a misinformation super-
spreader users classifier and a bot detector. It is im-
portant to highlight that the focus of this work is the
design of the Digital Lighthouse platform and the or-
chestration of its several components. For this reason,
we will not detail the algorithms, methods and strate-
gies used in the knowledge discovery. We will do this
in other papers.
1
https://www.whatsapp.com/legal/privacy-policy
3.4 Module IV: Data Visualization
Today, there is a great need for displaying massive
amounts of data in a way that is easily accessible and
understandable. In this context, data visualization is
a way to represent information graphically, highlight-
ing patterns and trends in data and helping to achieve
new insights. It enables the data exploration via the
manipulation of charts and images. More specifically,
it enables users to analyze the data by interacting di-
rectly with a visual representation of it. In this work,
the data visualization module is a web application
developed using Python programming language and
Django 3 framework.
4 CASE STUDY
To evaluate the platform proposed in this paper, we
performed an exploratory case study using three dif-
ferent WhatsApp’ messages datasets, covering rele-
vant themes such as the Brazilian general elections
campaign in 2018, the covid-19 pandemic the vac-
cine for covid-19. This case study was influenced
by (Jedlitschka and Pfahl, 2005; Kitchenham et al.,
2008; Robson and McCartan, 2016; Runeson and
H
¨
ost, 2009). Then, many data analysis techniques
were applied to this dataset to get insights about mis-
information spread.
Next, we will describe these three datasets in de-
tail.
Brazilian General Elections: This dataset contains
282,601 messages, obtained from 5,364 users
(cell phone chips), which participated in 59 What-
sApp public groups, in the period from August to
October 2018.
Covid-19 Pandemic: This dataset contains
228,061 messages, obtained from 10,495 users
(cell phone chips), which participated in 236
WhatsApp public groups, in the period from
March to June 2020.
Vaccine for Covid-19: This dataset contains
16,056 messages, obtained from 1,857 users (cell
phone chips), which participated in 175 What-
sApp public groups, in the period from December
2020 to January 2021.
Using the Data Visualization Module from the Light-
house Platform, the user can choose a specific dataset
or all data from all datasets. For simplicity, from this
point onwards, all graphs will be illustrated using the
Covid-19 dataset.
ICEIS 2021 - 23rd International Conference on Enterprise Information Systems
300
4.1 Messages Characterization
Initially, the Lighthouse Platform shows some visu-
alizations to characterize the used dataset. Figure
2 shows the proportion between messages with and
without URL. In general, messages created to spread
misinformation include URLs, often from a little-
known website or blog, to give credibility. Therefore,
the presence of a URL can be a criterion for select-
ing messages to be analyzed by fact-checkers. As you
can observe in Figure 2, a significant proportion of the
caught messages (9.33%) involves some URL.
Figure 2: Proportion between Messages with and without
URL.
Currently, audios, images, and videos are commonly
used to spread misinformation. Therefore, the mes-
sages associated with these files are potential can-
didates to undergo a verification process. Figure
3 shows the proportion between messages with and
without media. As you can note, 32.90% of the
caught messages involves some media file.
Figure 3: Proportion between Messages with and without
Media.
In the 2018 Brazilian elections, many cell phone chips
from foreign countries were used in the massive mes-
saging with an electoral advertisement. Thus, monitor
these messages is an important task to identify misin-
formation spreading. Figure 4 illustrates the propor-
tion of foreign countries messages.
Figure 5 shows the distribution messages send-
ing time by the day hours. As we can imagine, the
peak of sending messages occurs at the time reserved
Figure 4: Proportion of Foreign Countries Messages.
for lunch (between 12 and 14 hours) and in the early
evening, just after work hours.
Figure 5: Number of Messages by Hour.
4.2 Geographic Distribution
Another relevant aspect to observe in the monitored
groups is the geographic location of users (cell phone
chips), both Brazilians and foreigners, besides these
users’ activity level. Figure 6 shows the Brazilian
states with more quantity of messages. As might
be expected, the most populous states have the most
significant amount of messages sent. Figure 7 il-
lustrates the Brazilian states with more users’ (cell
phone chips). As might be expected, the most popu-
lous states have the most significant amount of users.
However, when analyzing the states with more mes-
sages per user (Figure 8), we can observe that not so
populous states such as Mato Grosso do Sul, Santa
Catarina, and Amazonas, have the most active users.
As previously mentioned, cell phone chips from
foreign countries have been used in Brazil for mas-
sive messaging, many times spreading misinforma-
tion. Figure 9 illustrates the number of messages
sent by foreign countries cell phone chips by country,
while Figure 10 shows the countries with the lagers
ratio between sent messages and the number of users.
Digital Lighthouse: A Platform for Monitoring Public Groups in WhatsApp
301
Figure 6: States with more Messages.
Figure 7: States with more Users.
Figure 8: States with more Messages per User.
Figure 9: Messages by Foreign Countries.
Figure 10: Countries with More Active Users.
4.3 Vocabulary Characterization
Another aspect that needs to be analyzed is related
to the characteristics of the vocabulary used in the
text messages, since there is a strong relationship be-
tween the used vocabulary and the social network, in
this case, WhatsApp. Figure 11 shows the number
of messages by the number of words contained in the
message. As we can note, there are few messages
with a large number of words and a high number of
messages with few words. Figure 12 shows the word
cloud highlighting the most popular words.
Figure 11: Number of Messages by the Number of Words
in the Message.
Figure 12: Word Cloud.
4.4 Misinformation Analysis
The last aspect to be explored is the misinformation
analysis. In this context, various information about
messages and users are explored to identify text mes-
sages containing misinformation and super-spreaders.
ICEIS 2021 - 23rd International Conference on Enterprise Information Systems
302
So, we first built a new dataset adding only the
messages with at least one of these words: ‘covid’,
’corona’, ’coronga’, or virus’. The resulting dataset
had 3,014 messages. Table 1 contains the seven most
shared messages. The “Sharings” column indicates
how many times the message was shared. The “Mis”
column indicates if the messages contains or not the
misinformation. Finally, the column “NoG” denotes
the number of distinct groups where the message was
shared. Note that all the seven most shared messages
contain misinformation.
Table 1: Most Shared Messages.
Sharings Text Mis NoG
43 ”PATRIOTA! *VAMOS ACOR-
DAR BRASIL!!!! E VOCE AINDA
ACREDITANDO NESTA FARSA DE
COVID19,
´
E UM GOLPE QUE FOI
ARQUITETADO PARA ENGANAR OS
BRASILEIROS, MENOS ESCLARECI-
DOS...* #NAOFIQUEEMCASA #VA-
MOSTRABALHAR #BOLSONAROES-
TACERTO
Yes 40
26 ”Pesquisa com mais de 6.000 m
´
edicos em 30
pa
´
ıses diz que hidroxicloroquina
´
e o trata-
mento mais eficaz para coronav
´
ırus.
Yes 23
23 ”Dra. Nise Yamaguchi integra gabinete de
crise e prop
˜
oe a cloroquina como tratamento
imediato nos casos de coronav
´
ırus.
Yes 23
23 ”Heranc¸a maldita: Mandetta renova contratos
de publicidade de R$ 1 bilh
˜
ao firmados no
governo Dilma...
Yes 14
22 ”Organizac¸
˜
ao Mundial de Sa
´
ude: O aborto
´
e “essencial” durante a pandemia de coro-
nav
´
ırus chin
ˆ
es.
Yes 22
18 ”Prezados amigos.. voc
ˆ
es sabiam que, todos
os problemas da humanidade foram curados
com esse p
ˆ
anico fake do covid19?????? Ve-
jam?? Sempre morreram milhares de pes-
soas de H1N1, POIS, NUNCA FOI ERRADI-
CADA ESTA GRIPE, DE AIDS que NUNCA
FOI ERRADICADA, DE TUBERCULOSE,
DE INFARTO, DE BRIGAS DOM
´
ESTICAS,
DE IDADE, DE INSUFICI
ˆ
ENCIA RESPI-
RAT
´
ORIA, DE C
ˆ
ANCER, DE DIVERSAS
OUTRAS DOENC¸ AS E MALES... TUDO
ACABOU...
Yes 18
16 ”*Atenc¸
˜
ao*: Isso a Globo n
˜
ao mostra. Banco
Mundial acaba de lanc¸ar um documento que
ressalta o papel do com
´
ercio internacional na
mitigac¸
˜
ao dos impactos do coronav
´
ırus. A
instituic¸
˜
ao argumenta que a manutenc¸
˜
ao dos
fluxos de com
´
ercio ser
´
a crucial para o supri-
mento de itens m
´
edicos e alimentos — e por-
tanto limitar impactos negativos sobre empre-
gos e n
´
ıvel de pobreza em escala global. O tra-
balho do Banco Mundial coloca o Brasil como
“Exemplo 1” no quadro “Melhores Pr
´
aticas
em Lidar com a Covid-19”. #BolsonaroTem-
Raz
˜
ao”
Yes 11
Table 2 contains the 5 most active users together
with the number of messages shared by each one.
The user identification was anonymized. Let’s take
a particular user, for example, the user with Id
-9126362355320474072, which sent 67 messages.
Table 3 contains all messages shared by the user
-9126362355320474072. Note that all 67 mes-
sages shared by this user have misinformation. Be-
sides, some messages were shared many times.
Now, let’s take a specific message of the user -
9126362355320474072, as, for example, the message
in the first row of Table 3. Table 4 contains the date
and time of each sharing of the selected message, be-
sides the group in which it was shared. Note that the
selected message was shared 22 times in 22 differ-
ent groups, in a period of four minutes. So, we can
classify the user -9126362355320474072 as a misin-
formation super-spreader.
Table 2: Most Active Users.
User Id Number of Messages
3346599479176653344 110
8121536360444460807 102
-9126362355320474072 67
8900877460624761918 62
1721737435325801397 60
Table 3: Messages of User Id -9126362355320474072.
Sharings Text Mis
22 ”Pesquisa com mais de 6.000 m
´
edicos em 30 pa
´
ıses
diz que hidroxicloroquina
´
e o tratamento mais efi-
caz para coronav
´
ırus.
Yes
22 ”Dra. Nise Yamaguchi integra gabinete de crise e
prop
˜
oe a cloroquina como tratamento imediato nos
casos de coronav
´
ırus.
Yes
22 ”Organizac¸
˜
ao Mundial de Sa
´
ude: O aborto
´
e
“essencial” durante a pandemia de coronav
´
ırus
chin
ˆ
es.
Yes
1 ”ENTENDA COMO FOMOS IMPEDIDOS DE
VOTAR O FUND
˜
AO PARA O COMBATE AO
CORONAV
´
IRUS...
Yes
5 CONCLUSIONS
The fast spread of misinformation through What-
sApp messages poses a significant social problem.
In this work, we present an entire platform, called
Digital Lighthouse, that aims for finding WhatsApp
public groups, besides extracting, cleaning, analyz-
ing, and visualizing misinformation that spread in
such groups. Using the proposed platform we build
three different WhatsApp’ messages datasets, cover-
ing relevant themes such as the Brazilian elections,
the covid-19 pandemic, and the vaccine for covid-19.
Besides, we presented a case study using the pro-
Digital Lighthouse: A Platform for Monitoring Public Groups in WhatsApp
303
Table 4: Details of the Selected Message.
Date Time Group Id
2020/04/06 18:36 2020 117
2020/04/06 18:36 2020 133
2020/04/06 18:36 2020 153
2020/04/06 18:36 2020 187
2020/04/06 18:36 2020 243
2020/04/06 18:36 2020 26
2020/04/06 18:36 2020 96
2020/04/06 18:37 2020 128
2020/04/06 18:37 2020 131
2020/04/06 18:37 2020 174
2020/04/06 18:37 2020 84
2020/04/06 18:38 2020 146
2020/04/06 18:38 2020 170
2020/04/06 18:38 2020 171
2020/04/06 18:38 2020 22
2020/04/06 18:38 2020 225
2020/04/06 18:38 2020 229
2020/04/06 18:38 2020 233
2020/04/06 18:38 2020 73
2020/04/06 18:38 2020 99
2020/04/06 18:39 2020 105
2020/04/06 18:39 2020 226
posed platform. Initially, we characterize the used
dataset, explored the geographic distribution of the
messages and performed a vocabulary characteriza-
tion. Finally, we performed a misinformation analy-
sis and we identified a misinformation super-spreader.
As future work we will extend the Lighthouse plat-
form using big data and real-time technologies.
REFERENCES
Faustini, P. and Cov
˜
oes, T. (2019). Fake news detection
using one-class classification. In 2019 8th Brazilian
Conference on Intelligent Systems (BRACIS), pages
592–597.
Gaglani, J., Gandhi, Y., Gogate, S., and Halbe, A. (2020).
Unsupervised whatsapp fake news detection using se-
mantic search. In 2020 4th International Conference
on Intelligent Computing and Control Systems (ICI-
CCS), pages 285–289. IEEE.
Garimella, K. and Tyson, G. (2018). Whatsapp, doc? a first
look at whatsapp public group data. arXiv preprint
arXiv:1804.01473.
Hamdi, T., Slimi, H., Bounhas, I., and Slimani, Y. (2020).
A hybrid approach for fake news detection in twitter
based on user features and graph embedding. In Hung,
D. V. and D’Souza, M., editors, Distributed Comput-
ing and Internet Technology - 16th International Con-
ference, ICDCIT 2020, Bhubaneswar, India, January
9-12, 2020, Proceedings, volume 11969 of Lecture
Notes in Computer Science, pages 266–280. Springer.
Jedlitschka, A. and Pfahl, D. (2005). Reporting guidelines
for controlled experiments in software engineering. In
Empirical Software Engineering, 2005. 2005 Interna-
tional Symposium on, pages 10–pp. IEEE.
Kitchenham, B., Al-Khilidar, H., Babar, M. A., Berry,
M., Cox, K., Keung, J., Kurniawati, F., Staples, M.,
Zhang, H., and Zhu, L. (2008). Evaluating guidelines
for reporting empirical software engineering studies.
Empirical Software Engineering, 13(1):97–121.
Machado, C., Kira, B., Narayanan, V., Kollanyi, B., and
Howard, P. (2019). A study of misinformation in
whatsapp groups with a focus on the brazilian presi-
dential elections. WWW ’19, page 1013–1019, New
York, NY, USA. Association for Computing Machin-
ery.
Qiu, X., Oliveira, D. F., Shirazi, A. S., Flammini, A., and
Menczer, F. (2017). Limited individual attention and
online virality of low-quality information. Nature Hu-
man Behaviour, 1(7):0132.
Resende, G., Melo, P., Sousa, H., Messias, J., Vascon-
celos, M., Almeida, J., and Benevenuto, F. (2019).
(mis)information dissemination in whatsapp: Gather-
ing, analyzing and countermeasures.
Resende, G., Messias, J., Silva, M., Almeida, J., Vascon-
celos, M., and Benevenuto, F. (2018). A system for
monitoring public political groups in whatsapp. In
Proceedings of the 24th Brazilian Symposium on Mul-
timedia and the Web, WebMedia ’18, page 387–390,
New York, NY, USA. Association for Computing Ma-
chinery.
Robson, C. and McCartan, K. (2016). Real world research.
Wiley.
Runeson, P. and H
¨
ost, M. (2009). Guidelines for conduct-
ing and reporting case study research in software engi-
neering. Empirical software engineering, 14(2):131–
164.
Shu, K., Bernard, H. R., and Liu, H. (2018). Studying fake
news via network analysis: Detection and mitigation.
CoRR, abs/1804.10233.
Shu, K., Zhou, X., Wang, S., Zafarani, R., and Liu, H.
(2019). The role of user profiles for fake news de-
tection. In Proceedings of the 2019 IEEE/ACM Inter-
national Conference on Advances in Social Networks
Analysis and Mining, ASONAM ’19, page 436–439,
New York, NY, USA. Association for Computing Ma-
chinery.
Silva, R. M., Santos, R. L., Almeida, T. A., and Pardo,
T. A. (2020). Towards automatically filtering fake
news in portuguese. Expert Systems with Applications,
146:113199.
Su, Q., Wan, M., Liu, X., and Huang, C.-R. (2020). Mo-
tivations, methods and metrics of misinformation de-
tection: An nlp perspective. Natural Language Pro-
cessing Research, 1:1–13.
Vosoughi, S., Roy, D., and Aral, S. (2018). The spread of
true and false news online. Science, 359:1146–1151.
Zhang, Y. and Hara, T. (2020). A probabilistic model for
malicious user and rumor detection on social media.
In 53rd Hawaii International Conference on System
Sciences, HICSS 2020, Maui, Hawaii, USA, January
7-10, 2020, pages 1–10. ScholarSpace.
ICEIS 2021 - 23rd International Conference on Enterprise Information Systems
304