Scientometric Analysis of Fake News Detection and Machine Learning

Based on VOSviewer

Lumbardha Hasimi and Aneta Poniszewska-Mara

nda

Institute of Information Technology, Lodz University of Technology, Ł

z, Poland

Keywords:

Fake News Detection, Machine Learning, Scientometrics, VOSviewer.

Abstract:

This study presents a comprehensive analysis of recent research patterns and progress in the ﬁeld of fake

news detection and machine learning. By examining 2209 publications from 2015 to 2022, the study aims

to identify the most fre-quently developed topics and explore the involvement of publications, authors, and

institutions. Using the network visualizing tool VOSviewer, a quantitative analysis is performed to investi-

gate research productivity, patterns, and keyword distribution. This study contributes to the understanding of

the current state of research in fake news detection and machine learning, and offers valuable in-sights for

researchers, policymakers, and technology developers seeking to ad-dress the challenges posed by fake news

and disinformation. The ﬁndings indi-cate that fake news detection research is still in its early stages and

primarily focuses on social media and social contexts. There is a growing interest in the subject, as evidenced

by increasing attention from the research community, whereas the network of interconnected research clusters,

highlights the multidis-ciplinary nature of fake news detection.

1 INTRODUCTION

The spreading potential of fake news has become one

of the biggest challenges in recent years. The massive

dissemination of information has emerged as a fright-

ening issue worldwide. The prevalence of fake news

has been present for a long period, however with the

advancements of mass media, it has come to be one of

the biggest concerns of the online world. According

to new Central Statistics Ofﬁce (CSO) data, almost

two-thirds of Internet users have been exposed to on-

line content they considered to be not true or doubtful

in 2021 (CSO, 2020). Fake news creates an adverse

impact in every area, be it defaming, changing pub-

lic opinion on political opinion, or simply ﬁnancial,

entertainment, and/or personal gain (Choras, 2020).

Initiatives of considerable signiﬁcance have

started from a worldwide perspective. The Interna-

tional Grand Committee (IGC) on Disinformation and

Fake News is among the created boards focused on

technology and media companies, and accountability

in fake news issues (Tavares et al., 2017). The prob-

lem of fake news has grown into a major challenge

for many societies. This phenomenon reaches poli-

tics, organizations, and individuals having an impact

https://orcid.org/0000-0001-7596-0813

in different spheres. The most recent example of the

proliferation and risk of fake news dissemination is

the spread of anti-vaccination misinformation or the

rumours regarding the incorrectly compared number

of registered voters in 2018 to the number of votes

cast in US Elections 2020 (Reuters, 2021). Its preva-

lence has shown certain patterns, especially during

certain periods. For instance, elections, outbreaks etc.

Therefore, it is critically important to stop the spread

of fake news at an early stage.

However, with the ongoing technological ad-

vancements, the format of fake news is advancing as

well. In recent times, the detection of multimodal for-

mat is becoming an issue on the rise. Visual and video

propagation are new domains concerning the research

community (Rohman et al., 2021; Jain and Kasbe,

2018). False information also attracts the attention

of academia from various disciplines. Current knowl-

edge bases struggle to validate false news effectively

when it is linked to time-critical events as there is a

lack of supporting claims or facts (Vinhas and Bastos,

2022). Furthermore, the nature of the data and the

structure of the raw fake information, does not fol-

low a particular pattern. Researchers have attempted

in recent years to uncover problems with false news

and offer solutions, especially regarding social media

and dissemination. Nevertheless, according to (Paor

548

Hasimi, L. and Poniszewska-MaraÅ

Dda, A.

Scientometric Analysis of Fake News Detection and Machine Learning Based on VOSviewer.

DOI: 10.5220/0012128300003538

In Proceedings of the 18th International Conference on Software Technologies (ICSOFT 2023), pages 548-555

ISBN: 978-989-758-665-1; ISSN: 2184-2833

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

and Heravi, 2020) not only the online solutions are a

tool to ﬁght fake news, the literacy and education are

also essential in combating the spreading of false in-

formation. The academic community has risen up to

the challenge, investigating the causes, the scope and

scale, the detection method, and how they spread to

avoid their dramatic impacts (Wang, 2020; Gerbina,

2021).

With the rapid increase in the spread of fake news,

the research community most prominently observed

AI use in the cause against fake news dissemination.

Particularly, the use of machine learning as a promis-

ing solution, especially regarding the real-time fac-

tor. Machine learning can be used to track and anal-

yse information that can be validated by a legitimate

medium and this way automate the review process,

without human intervention (Alharbi et al., 2021). Al-

though it is difﬁcult to prevent the creation and spread

of fake news articles, machine learning algorithms

can be employed to detect anomalies and patterns

(Khalil et al., 2021; Agrawal et al., 2021) that can lead

to the prevention of fake news spreading. The use of

a machine learning approach against fake news and

disinformation can not only increase the efﬁciency

against the scalability of fake news but also speed up

the prevention/detection process. Many researchers

have tackled the issue of fake news detection in terms

of machine learning methods (Tavares et al., 2017;

Khalil et al., 2021; Agrawal et al., 2021; Biwalkar

et al., 2021; Abdulrahman and Baykara, 2020; Babu

et al., 2022). To analyse further such contributions,

through the scientometric analysis method it is possi-

ble to investigate the quantity and quality of research

on this topic. Articles related to fake news detection

have attracted interest, and given the numbers, last

ﬁve years the topic of fake news detection has shown

a rapid increase. Thus, this study aims to delve into

existing sources in order to provide a comprehensive

and objective analysis of fake news detection, under

different criteria and analysis.

The objective of this study is to ultimately help

in the selection and identiﬁcation of core literature in

the ﬁeld, the latest trends, and developments in fake

news detection, while comprehensively investigating

the most recent advancements in academia and the re-

search community.

The paper is structured as follows: Section 2

presents the overview of methodology used for the

conducted research and experiments. Section 3 de-

scribe the details of analysis of publishing patterns,

while section 4 deals with the research productivity

and research impact in the area of fake news and dis-

semination.

Table 1: PICO strategy – Keywords and synonyms.

Keyword Synonyms Related to

detection classiﬁcation Intervention

identiﬁcation

veriﬁcation

fake news disinformation Population

fake information

false news

misinformation

machine

learning

supervised methods, unsupervised

methods, reinforcement learning

Intervention

2 METHODOLOGY

The study utilizes data from the Scopus dataset as one

of the largest scientiﬁc databases of multi-disciplinary

publications (Schotten et al., 2017). The collection

of relevant publications and citations establishes the

foundation for a scientometric analysis of a speciﬁc

research area (Mazov et al., 2020; Khokhlov, 2020).

This study covers a big number of peer-reviewed ar-

ticles published in the last six years. Through this,

we seek to attain evidence regarding fake news de-

tection and machine learning. To achieve the objec-

tive of this study we processed indexed publications

of the highest quality. For the scientometric study,

formalized keywords are of crucial importance in the

process of data collection. After the formalization of

keywords, we designed and ran the following search

query: TITLE-ABS-KEY ( ( ”fake news” OR ”disin-

formation” OR ”fake information” OR ”false news”

OR ”misinformation” ) AND ( ”detection” OR ”clas-

siﬁcation” OR ”identiﬁcation” OR ”veriﬁcation” ) )

AND ( ”machine learning” ) AND PUBYEAR ¿ 2014

AND PUBYEAR ¿ 2014. The ﬁnal query was ob-

tained considering the keywords and synonyms using

PICO model – search strategy presented in table 1.

Considering that the collection of relevant pub-

lications and citations is the foundation for a sci-

entometric analysis, we gathered data for a speciﬁc

span of years, from 2015 to 2022. All the available

results from the executed query had been exported

and recorded as a dataset ﬁle. The data visualiza-

tion constructions, network assessments, and cluster-

ing performed are conducted with the support of the

VOSviewer program (van Eck and Waltman, 2010;

(SMU), 2020). VOSviewer is developed as a power-

ful software for constructing and visualizing biblio-

metric networks and keyword occurrences, allowing

manipulations with a large number of data extracted

from some of the most-known databases (van Eck and

Waltman, 2010). The program allows data visualiza-

tion that makes it possible to have a fully detailed ex-

amination of speciﬁc data. To achieve a fuller and

Scientometric Analysis of Fake News Detection and Machine Learning Based on VOSviewer

549

Table 2: Results of number of publications per year.

Year Documents RGR

2022 606 0.11

2021 685 0.66

2020 352 0.49

2019 215 0.89

2018 88 0.84

2107 38 1.24

2016 11 0.32

2015 8 -

more complete map it was aimed in getting as greater

number of connections as possible. To ensure consis-

tent analysis, we focused on parameters such as the

type of publication, citation patterns per author and

institution, research productivity, and keyword analy-

sis.

2.1 Data Processing

The data collected from Scopus resulted in 2017 pub-

lications. Besides the query parameters, no other ﬁl-

ters were applied, as the quantity of the publications

is sufﬁcient for the scientometric analysis. The scien-

tiﬁc input in the database through the observed years

is presented in table 2. As seen from the results, the

highest increase in publications. Moreover, Scopus

with the new version offers even more advanced func-

tionality to export structured data. In order to investi-

gate the research productivity, several attributes of the

collected data were utilized to analyse various aspects

of publications.

To ensure the data is all accurate with no informa-

tion missing, the ﬁnal number of publications retained

2003, containing 7413 keywords in all keyword’s unit

analysis. The most productive year in terms of impact

and the number of papers published, resulting also in

the highest relative growth rate (Baskaran, 2022) from

year to year, is 2021 with almost double the number

of records compared to 2020 (Equation 1).

Equation 1. The calculation of relative growth

rate

RGR = (lnN

− lnn

)/(t

− T

)

Selection of inappropriate form of publication has

an inﬂuence in the visibility of research, hence its im-

pact as well. For this reason, we decided to analyse

which type of publication venue, the majority of the

researchers in this study convey their insights.

Table 3 indicates that conference papers make

the most of the document types, with an average of

57.8%, whereas an average of 98.9 percent of these

studies are written in the English language. In this

surely has an inﬂuence on the fact the majority of re-

search is from the Computer Science discipline, and

as well the most dominant having conference papers

Table 3: Publications and language.

Type Percentage

Document

type

Conference Paper (57.9%), Article (32%),

Book Chapter (3.2%), Conference Re-

view (3.3%), Review (2.2%), Book (0.1%),

Data Paper (0.1%), Letter (0.1%), Re-

tracted, Editorial, Note, Short Survey, oth-

ers (0.1%)

Language English (98.9%), Chinese (0.4%), Por-

tuguese (0.4%), Others (0.3%)

as the primary means of publication (Heilig and Vob,

2014). Nevertheless, another explanation lies in the

fact that the topic is quite novice, and the timely pre-

sentation is important to consider, especially when

dealing with a rapidly growing issue.

2.2 Research Patterns and Units

In order to measure the research impact, we imple-

mented citation analyses, speciﬁcally the number of

current citations per author, document, and institu-

tion. The scientometric approach of the article is de-

signed to explore such interactions at the levels of

topics, publication venues, disciplines, and institu-

tions. Similarly, co-citation analyses were also exe-

cuted, in order to observe a relational dimension of

the research network. This resulted in the formation

of connections between authors, articles, and institu-

tions, which are the foundations of this study. The

option of citation-based clustering in VOSviewer of-

fers visualised groups of data that share signiﬁcant

similarities. This helps to draw connections and dif-

ferentiations of the data into separate categories. To

have even more comprehensive overview, the overlay

option builds maps representing the timeline with dif-

ferent colours. In particular, this approach allows to

identiﬁcation of research fronts based on relationships

between the data and follows the evolution of the re-

search by means of spatial connotations such as dis-

cipline, ﬁelds and other parameters, visualized across

time (van Eck and Waltman, 2010).

Keyword Analysis and Other Relevant Aspects

To further explore key topics and aspects of fake

news detection and machine learning we implemented

the co-occurrence analyses, in order to build key-

word clusters and observe the most frequent keywords

against intersections with other ﬁelds. To clearly ob-

serve the spectrum of keywords, we used full calcula-

tion methods to obtain a theme map of author key-

words and all keywords. The data were extracted

from the title and abstract and processed using the full

counting method.

ICSOFT 2023 - 18th International Conference on Software Technologies

550

3 ANALYSIS OF PUBLISHING

PATTERNS

To have a wider view in the basic structure of fake

news in machine learning research we analysed the

data from different perspectives. The analysis con-

sists of a distribution of the involved research disci-

plines, the contributing institutions, the number of au-

thors and the distribution of documents.

3.1 Academic Disciplines

We start with analyses of academic disciplines in or-

der to obtain an understanding of the general structure

and the development of the subject. The distribution

of publications for the entire period of six years is pre-

sented in ﬁgure 1.

It is evident the contribution in the Computer Sci-

ence subject is nearly constant throughout the years,

whereas the number of contributions in other disci-

plines slightly changes, especially last three years.

This indicates that the effects of fake news detection,

and its intersection with other disciplines is still at a

developing point.

The results reveal the strict inclusion of three main

subjects in the topic of fake news detection, namely

Computer Science, Engineering, and Mathematics.

However, in 2021 there was a slight trend of Decision

Sciences to take over Mathematics, especially if such

results are compared to 2015 and 1016. In particular,

the peak popularity of the subject is distinct during the

Covid-19 pandemic, which had a huge impact on the

research output (Raynaud, 2021). According to the

search result on the title and abstract content, out of

the whole number, 153 of the publications are Covid-

19 related to fake news, dissemination, datasets, etc.

This implies that research activities during this period

were partially affected by the ongoing situation.

To further evaluate research productivity, it is cru-

cial to identify the most active research institutes in

the ﬁeld. Such insight is useful in building research

collaborations and reﬂecting on a global scale con-

cerning the distribution of research (Srainternational,

2020).

Figure 2 shows the rankings of research institutes

ordered by the number of publications. Evidently,

the numbers demonstrate a dominance of publications

from Indian institutions. Delhi Technological Univer-

sity, as one of the most reputable institutions in India,

has the highest number of publications. Followed by

Arizona State University and the Chinese Academy

of Sciences. Most of these institutions have an ex-

cellent reputation in research and attract some of the

best scientists in the ﬁeld with broad knowledge and

expertise.

In order to observe the research productivity,

while considering the limitations of the methods used,

we focus on the number of papers per author, along-

side citation patterns and numbers.

3.2 Co-Authorship Distribution

Analysis

To obtain a deeper insight into contribution patterns,

we further investigate the distribution of publications

per citation and authorship patterns. The majority of

the research on fake news detection is carried out by

researchers from India, given the individual produc-

tivity.

To provide sufﬁcient insight into the relevance of

the contributions we also run the co-authorships anal-

ysis, alongside the citation analysis – presented in the

next section. We analysed co-authorship with units of

analysis of authors using the full counting method to

get the authors’ collaboration network.

Figure 3 shows 11 clusters of constructed patterns

in the authors’ collaboration network. The criteria of

a minimum number of documents per author was set

to 2, and to ensure the selection of high-impact data,

the number of citations per author was set to 50. Thus,

out of 4798 authors, only 74 are presented in the map

creation. The largest clusters of collaboration patterns

are between researchers of clusters 1 and 2, both con-

taining 13 items. As seen from the ﬁgure, inter-cluster

collaboration is more common among the leading au-

thors. These authors have collaboration in wide as-

pects of fake news detection and machine learning.

Some of the authors with various collaborations are

Nakov, P., Shu, K., Liu, H., Da San Martino, G.,

Alam, F. Collaborative work is very important, espe-

cially in the case of an issue with the interdisciplinary

outcome. Hence, through collaborative work not only

productivity is affected, but also author and publica-

tion visibility is inﬂuenced by positive network mem-

bership, given the case of inﬂuential outlets.

3.3 Frequent Keywords and Keyword

Clusters; Co-occurrence Network of

High-Frequency Keywords

Keywords indicate the core ﬁelds of concern and

represent an effective instrument in the classiﬁca-

tion of the content of scientiﬁc work. From a

meta-perspective, keywords are the foundation for

analysing the key topics and aspects representing a

particular research area (Heilig and Vob, 2014). Co-

occurrence analysis of keywords, not only help in

Scientometric Analysis of Fake News Detection and Machine Learning Based on VOSviewer

551

Figure 1: Distribution of documents per discipline.

Figure 2: Distribution of documents per afﬁliation.

Figure 3: Co-authorship analysis map per author unit.

quick identiﬁcation of popular topic within a time-

frame but also help in pointing out aspects and topics

related to each other. For this reason, we decided to

observe closely the distribution of keywords, in both

units: all are indexed (Fig. 4).

The observed papers comprising research related

to fake news detection machine learning provide 7413

keywords in total. The most often cited expressions

are: ”social networking (online)” (711), ”fake de-

tection” (711), ”fake news” (697), ”social media”

(613), ”fake news detection” (474), ”machine learn-

ing” (411). According to (Guo et al., 2017), using

the relevant formula, the number of high-frequency

keywords recommended for further analysis of co-

occurrence should include 100 high-frequency key-

words. On the other hand, for the 7 clusters ob-

tained out of 314 words, as classiﬁed by VOSviewer,

have the following dominant key-words per cluster:

the ﬁrst cluster shows the foremost keyword ”social

network online” and ”fake news”, the second cluster

”fake news detection”, the third cluster ”fake detec-

tion”, the fourth cluster ”natural language process-

ing”, the ﬁfth cluster ”social media”, the sixth clus-

ter ”embeddings”, the seventh cluster ”machine learn-

ing”. The results of the keyword analysis further re-

vealed that the there is a tendency of approaching fake

news detection mainly through the lenses of social

media and social context.

Nevertheless, using VOSviewer, out of the to-

tal number of keywords, only 314 met the threshold

of the minimum occurrences per keyword set to 10,

while being classiﬁed in 7 clusters. In the map pre-

ICSOFT 2023 - 18th International Conference on Software Technologies

552

Figure 4: Co-occurrence analysis – all keywords.

sented, the size of nodes manifests the frequency of

keyword’s occurrence, while lines show relationships

among keywords (Table 4).

4 RESEARCH PRODUCTIVITY

AND RESEARCH IMPACT

Having the numbers that provide insights into pub-

lishing patterns alone is not sufﬁcient to clarify the

impact of research. One of the primary concerns of

a scientometric study is to assess the impact of such

contributions. A measure for analysing the impact of

contributions is the aggregated number of citations a

publication receives (Heilig and Vob, 2014). To mea-

sure the research impact, we applied citation and co-

citation analysis at individual and institutional level

(Fig. 5).

For citation analysis using the author unit, out of

4798, 64 meet the threshold. To create the map of

citations per author, the minimum number of doc-

uments per author was set to 5, and the minimum

number of citations per an author was set to 50. As

a result, a network of 58 items (the largest set of

connected items) was created, consisting of 8 clus-

ters (Fig. 5). The biggest cluster, consists of 12 au-

thors, depicted in red colour, and shows the network

of some of the most productive authors such as Liu,

H., Sharma D.K., Shu, K.

To receive the co-citation analysis of institutions,

the minimum contribution was set to 1, and the min-

imum number of citations to 50. In return, 56 items

were mapped (Fig. 6). Out of the overall number of

items, 9 clusters were obtained, with the ﬁrst cluster

being the largest – consisting of 12 items, whereas the

smallest the last one consisting of 3 items.

Reference co-citation analysis is one of the most

important tools to analyse and reﬂect the evolutionary

process in a particular scientiﬁc activity (Ding et al.,

2021). We received overall 57, 355 cited references,

for the co-citation network, out of which only 26 are

represented below in the network, following the 20

citations per reference criteria. Figure 7 shows the

mapping on the co-citation of references.

The red colour cluster indicates the ﬁrst cluster,

which through network lines identiﬁes the references

cited in the common paper. The frequency is depicted

through the size of the dot – in this case, the reference

point. From this network, 4 clusters were acquired,

having the ﬁrst cluster consisting of 9 items, and the

smallest cluster 4, consisting of 5 items. The ﬁrst

cluster encompasses research regarding techniques,

perspectives, and methods of fake news and its de-

tection. The second cluster concentrates on fake news

detection in social media and its detection within the

social context. The third cluster investigates infor-

mation credibility and fake news spreading patterns.

The fourth cluster focuses on the machine learning

approach, results, and efﬁciency of models. These re-

sults indicate that there is rapidly growing research on

the topic of fake news detection and machine learn-

ing, especially in terms of inter-institutional and inter-

disciplinary collaboration.

5 CONCLUSIONS

The latest trends and developments in fake news de-

tection have shown increasing attention from the re-

search community. The subject is gaining a lot of at-

tention from an interdisciplinary scope, posing a chal-

lenge to many research ﬁelds. In this study, we con-

duct a scientometric analysis to comprehensively in-

vestigate the trends and developments in fake news

detection and machine learning literature. To analyse

publication patterns, research productivity, and iden-

tify various sources while investigating the biggest

contributions per author and institution, we conducted

a quantitative analysis using the network visualizing

tool VOSviewer. The results of the study revealed

that the research is currently dominated by computer

science and conveyed especially through conference

proceedings. The research activity is mainly inﬂu-

enced by highly recognised scientists and publica-

tions, carried out by some of the most reputable in-

stitutions worldwide. Concerning keyword analyses

it was drawn to conclusion that the current focus of

fake news detection lies mainly on social media and

social contexts of fake news detection.

Furthermore, intersections of fake news detection

and machine learning have shown to be cutting-edge

Scientometric Analysis of Fake News Detection and Machine Learning Based on VOSviewer

553

Table 4: Keywords analysis results – author vs. all keywords.

All keywords No. Link Author keywords No. Link

1. fake news 685 1282 fake detection 711 10377

2. fake news detection 474 765 social networking (online) 711 10641

3. machine learning 308 766 fake news 697 8932

4. deep learning 261 680 social media 613 8720

5. social media 243 557 fake news detection 474 6098

6. natural language processing 197 512 deep learning 411 5803

7. misinformation 119 307 machine learning 410 5685

8. covid-19 117 326 classiﬁcation (of information) 296 4784

9. twitter 80 218 natural language processing 241 3423

10. disinformation 72 191 natural language processing systems 235 3622

11. classiﬁcation 62 147 learning systems 198 3270

12. bert 57 149 learning algorithms 196 3183

Figure 5: The most cited authors – citation per author.

Figure 6: Citation per institution unit analysis.

Figure 7: Co-citations per reference.

alongside social media, as recent research depicts.

Based on the results, the article concludes with some

general remarks on research productivity, research

patterns, and keyword distribution. Future work can

further explore certain aspects and concepts of fake

news detection separately from social context and so-

cial media. While technological solutions are impor-

tant, education and media literacy play a crucial role

in combating the spread of fake news. Future research

should focus on developing effective educational in-

terventions, media literacy programs, and strategies to

promote critical thinking and information evaluation

skills. Studies related to fake news detection and ma-

chine learning are relatively new, and there is still a

ICSOFT 2023 - 18th International Conference on Software Technologies

554

huge gap related to the fake news detection issue and

cross-disciplinary concepts related to it that can be in-

vestigated in the future.

ACKNOWLEDGEMENTS

The publication was created in the framework of Pol-

ish National Agency for Academic Exchange under

the ”STER Programme – Internationalisation of Doc-

toral Schools” as part of the project ”Curriculum for

advanced doctoral education & training – CADET

Academy of Lodz University of Technology”.

REFERENCES

Abdulrahman, A. and Baykara, M. (2020). Fake news de-

tection using machine learning and deep learning al-

gorithms. In 2020 International Conference on Ad-

vanced Science and Engineering (ICOASE), Duhok,

Iraq, pages 18–23.

Agrawal, C., Pandey, A., and Goyal, S. (2021). A survey on

role of machine learning and nlp in fake news detec-

tion on social media. In 4th International Conference

on Computing, Power and Communication Technolo-

gies (GUCON), pages 1–7.

Alharbi, R., Vu, M., and Thai, M. (2021). Evaluating

fake news detection models from explainable machine

learning perspectives. In Proceedings of IEEE IN-

TERNATIONAL CONFERENCE ON COMMUNICA-

TIONS (ICC 2021).

Babu, D., Sushmitha, G., Lasya, D., Krishna, D., and Ra-

jesh, V. (2022). Identifying fake news using machine

learning. In 2022 International Conference on Elec-

tronics and Renewable Systems (ICEARS), pages 1–6.

Baskaran, S. (2022). Publications trends in big data: A sci-

entometric analysis.

Biwalkar, A., Rao, A., and Shah, K. (2021). Real or fake:

An intrinsic analysis using supervised machine learn-

ing algorithms. In 5th International Conference on

I-SMAC (IoT in Social, Mobile, Analytics and Cloud)

(I-SMAC), pages 372–380.

Choras, M. (2020). Advanced machine learning techniques

for fake news (online disinformation) detection.

CSO, C. S. O. (2020). Information Integrity. Online,

https://www.cso.ie/en/releasesandpublications/ep/p-

isshisi/householdinternetsecurityandinformationintegrity

2021/informationintegrity/.

Ding, Y., Wang, Y., and Wang, Y. (2021). It’s time to con-

front fake news and rumors on social media: A bib-

liometric study based on vosviewer. In 4th Intern.

Conference on Computer and Communication Engi-

neering Technology, pages 226–232.

Gerbina, T. (2021). Science disinformation: On the problem

of fake news.

Guo, D., Chen, H., Long, R., Lu, H., and Long, Q. (2017).

A co-word analysis of organizational constraints for

maintaining sustainability.

Heilig, L. and Vob, S. (2014). A scientometric analysis of

cloud computing literature.

Jain, A. and Kasbe, A. (2018). Fake news detection.

Khalil, A., Jarrah, M., Aldwairi, M., and Jararweh, Y.

(2021). Detecting arabic fake news using machine

learning. In 2nd International Conference on Intelli-

gent Data Science Technologies and Applications (ID-

STA), pages 171–177.

Khokhlov, A. (2020). How scientometrics became the most

important science for researchers of all specialties.

Mazov, N., Gureev, V., and Glinskikh, V. (2020). The

methodological basis of deﬁning research trends and

fronts.

Paor, S. D. and Heravi, B. (2020). Information literacy and

fake news: How the ﬁeld of librarianship can help

combat the epidemic of fake news.

Raynaud, M. (2021). Impact of the covid-19 pandemic on

publication dynamics and non-covid-19 research pro-

duction.

Reuters (2021). Fact check: ’133 million registered

voters’ argument uses ﬂawed logic. Online,

https://www.reuters.com/article/uk-factcheck-voters-

133-million-idUSKBN296284.

Rohman, M., Khairani, D., Hulliyah, K., Arini, P.,

Riswandi, and Lakoni, I. (2021). Systematic liter-

ature review on methods used in classiﬁcation and

fake news detection in indonesian. In 9th Interna-

tional Conference on Cyber and IT Service Manage-

ment (CITSM), pages 1–4.

Schotten, M., el Aisati, M., Meester, W., Steiginga, S.,

and Ross, C. (2017). A brief history of scopus: The

world’s largest abstract and citation database of scien-

tiﬁc literature.

(SMU), S. M. U. (2020). Using VOSviewer as a

bibliometric mapping or analysis tool in Busi-

ness, Management & accounting. Online,

https://library.smu.edu.sg/topics-insights/using-

vosviewer-bibliometric-mapping-or-analysis-tool-

business-management.

Srainternational (2020). Enabling Collaborative Work

in Higher Education: An Exploration of Enhanc-

ing Research Collaborations Within an Institution.

Online, https://www.srainternational.org/blogs/srai-

jra1/2019/12/08/enabling-collaborative-work-in-

higher-education.

Tavares, B., da Silva, E. S., and de Souza, A. D. (2017).

Risk management analysis in scrum software projects.

van Eck, N. and Waltman, L. (2010). Software survey:

Vosviewer, a computer program for bibliometric map-

ping.

Vinhas, O. and Bastos, M. (2022). Fact-checking misinfor-

mation: Eight notes on consensus reality.

Wang, C.-C. (2020). Fake news and related concepts: Deﬁ-

nitions and recent research developments.

Scientometric Analysis of Fake News Detection and Machine Learning Based on VOSviewer

555