Altoe Et Al., 2024

Hindawi
Human Behavior and Emerging Technologies

Volume 2024, Article ID 1069670, 20 pages
https://doi.org/10.1155/2024/1069670
Review Article
Online Fake News Opinion Spread and Belief Change: A
Systematic Review
Filipe Altoe ,1 Catarina Moreira ,2 H. Sofia Pinto ,1 and Joaquim A. Jorge 1
1
INESC-ID/DEI-Instituto Superior Técnico-Universidade de Lisboa, Lisbon, Portugal
2
Human Technology Institute, University of Technology, Sydney, Australia
Correspondence should be addressed to Filipe Altoe; [email protected]
Received 19 September 2023; Revised 22 February 2024; Accepted 21 March 2024; Published 30 April 2024
Academic Editor: Mirko Duradoni
Copyright © 2024 Filipe Altoe et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Fake news has been linked to the rise of psychological disorders, the increased disbelief in science, and the erosion of democracy
and freedom of speech. Online social networks are arguably the main vehicle of fake news spread. Educating online users with
explanations is one way of preventing this spread. Understanding how online belief is formed and changed may offer a
roadmap for such education. The literature includes surveys addressing online opinion formation and polarization; however,
they usually address a single domain, such as politics, online marketing, health, and education, and do not make online belief
change their primary focus. Unlike other studies, this work is the first to present a cross-domain systematic literature review of
user studies, methodologies, and opinion model dimensions. It also includes the orthogonal polarization dimension, focusing
on online belief change. We include peer-reviewed works published in 2020 and later found in four relevant scientific
databases, excluding theoretical publications that did not offer validation through dataset experimentation or simulation.
Bibliometric networks were constructed for better visualization, leading to the organization of the papers that passed the
review criteria into a comprehensive taxonomy. Our findings show that a person’s individuality is the most significant
influential force in online belief change. We show that online arguments that balance facts with emotionally evoking content
are more efficient in changing their beliefs. Polarization was shown to be cross-correlated among multiple subjects, with
politics being the central polarization pole. Polarized online networks start as networks with high opinion segregation, evolve
into subnetworks of consensus, and achieve polarization around social network influencers. Trust in the information source
was demonstrated to be the chief psychological construct that drives online users to polarization. This shows that changing the
beliefs of influencers may create a positive snowball effect in changing the beliefs of polarized online social network users.
These findings lay the groundwork for further research on using personalized explanations to reduce the harmful effects of
online fake news on social networks.
Keywords: fake news; influencers; online belief change; online opinion formation; polarization; social network sites
1. Introduction artifacts, increasing the difficulty for news consumers to

discern disinformation from real news. Generative AI is
Fake news has been linked to the rise of psychological disor- increasingly being used with malicious intent. There is evi-
ders [1], the increased disbelief in science [2], and the ero- dence of scammers producing AI-generated fake voice
sion of democracy and freedom of speech [3]. Americans messages to victims’ family members, giving them dire
believe fake news to be a more severe problem than most news of being in immediate danger and requesting mone-
of the other critical issues in the country [4]. Lately, deepfake tary help [7].
technology [5] has been used in the production of videos of For these reasons, fake news is a hot research subject at
artificial intelligence (AI)–created avatars impersonating the time of this writing [8]. The state-of-the-art is mainly
news anchors reporting on fake news [6]. The advancement targeting their detection. Initial approaches attempted to
of deepfake technology is adding to the realism of fake news use machine learning classifiers over news content and
hbet, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1155/2024/1069670 by CAPES, Wiley Online Library on [25/09/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 Human Behavior and Emerging Technologies
analyze the news source. However, these methods fell out of ative; and Press, or what about the cultural tendencies of the
favor due to the need for a level of manual annotation that environment drive a given work to be deemed creative.
renders the approaches unusable in practical settings. More- The product vantage point is of interest when the goal is
over, as the fake news spread in online platforms sometimes to produce something useful to humans [27], applicable to
tends to be of viral velocity [9], these classifiers are not well the fake news explanation use case. The literature offers sev-
suited to the task. The community identified this problem eral examples of computer-generated creative artifacts, such
and has started automatically adopting deep learning tech- as music parodies [28], memes [29], anecdotes, poetry [30],
niques to extract features from online news posts. However, and jokes [31]. More recently, the introduction of large lan-
these black box models need more transparency to earn the guage models (LLMs) [32] opened up a new realm of possi-
news consumer’s trust in their generated outputs. Detection bilities around AI-generated creative artifacts. LLM fine-
explainability and visualization research started to gain tuning is being researched as an approach to specialize LLMs
momentum [10–12], leveraging explainable AI (XAI) [13]. in specific tasks that are better aligned with human-
Even though research on XAI has evolved significantly, it generated tasks [33]. Emotion-based personalized explana-
is still producing explanations that are too technical and tions can leverage this research to present explanations in a
suited to machine learning experts rather than the general manner that best aligns with the user preferences and max-
audience [14]. imizes the value of the experience through emotion evoca-
More recently, the community seems to be developing tion. It has been shown that computers can generate
an understanding that fake news online consumers’ educa- artifacts that create an emotional impact [34]. Our longer-
tion may be as crucial as fake news detection [15]. In this term research objective is to use CC-generated fake news
context, education may require changing their opinions explanations to verify our general hypothesis. Figure 1 illus-
and their preconceived beliefs. Cold factual explanations trates the concept. Understanding the current approaches
defending an opposite position can sometimes backfire and used in explanations to educate fake news readers and how
further entrench news consumers’ preconceived beliefs, online belief evolves, in general, will provide insight into
especially in online social network site (SNS) groups where the research opportunities for validating this idea.
polarization is prevalent [16]. In response to this, the The remainder of the paper is organized as follows. Sec-
research community has started to explore approaches for tion 2 highlights the intersections and differences in the cur-
changing consumers’ nonfactual beliefs [12] that could be rent literature review works to our proposed scope. Section 3
more efficient alternatives to the usual human-generated presents and explains the research questions that motivate
fact-checked explanation articles [17]. In [18], the authors the review methodology. The methodology itself is included
argue that explanations that nudge readers into a reflective in Section 4. This section presents separate discussions on
state are more efficient than purely factual ones in changing the rationale behind the inclusion and exclusion criteria,
user beliefs of fake photographs. Analyzing emotions in the search methodology applied, and a taxonomy organizing
speeches and consequent explanations using contrastive ele- the included works in clusters to facilitate analysis. Section 5
ments has also been investigated [19]. This motivated presents the findings for each review domain. Section 6 dis-
researchers to explore novel methods for creating more cusses the reviewed papers, highlighting how the findings
nuanced and emotionally resonant explanations to encour- address the proposed research questions and present other
age people to reflect on their beliefs. Some studies have pro- identified patterns. This section also presents the identified
posed generating artistic or emotional explanations as an grand challenges and related future work. Conclusions are
alternative approach to changing user beliefs rather than summarized in Section 7.
relying solely on facts [20, 21]. These explanations are aimed
at evoking an emotional response from the newsreader and 2. Related Work
gently nudging them into a reflective state.
Another work proposed a roadmap to personalizing fake To the best of our knowledge, this is the first cross-domain
news explanation systems [22]. Our general hypothesis is systematic literature review that includes the models, meth-
that fake news explanations personalized to some level and odologies, and user studies focusing on their influence on
evoke an emotional response carry better odds of changing belief formation and how preconceived beliefs are changed
users’ preconceived beliefs than purely factual explanations. in online and social network platforms. Furthermore, we
pay special attention to the role that polarization, an orthog-
1.1. Computational Creativity (CC). CC is a subfield of AI onal dimension to the three chief ones, plays in online belief
research that focuses on computational systems that exhibit change. This section reviews related work on each domain
behaviors that unbiased observers could deem creative [23]. and highlights the current gaps our work fills.
One of the most popular CC theories offered by the litera-
ture [24] has its foundations in what is known as the four 2.1. Models. From a model’s perspective, modeling of opin-
Ps [25, 26]. ion dynamics has been an active object of study [35], driving
The four P’s theory categorizes the study of creativity the community to generate surveys of online opinion prop-
from four vantage points: Person, or what about the agent agation [36] and trust propagation [37]. The author in [36]
makes them creative; Process, or what sort of actions are per- includes topics of interest to our work, such as stubborn
formed in the manufacturing of creative work; Product, or agents, biased agents, and opinion manipulation, the work
what about the output artifact is worthy of being called cre- reviewed papers published before 2019. From a model’s
Human Behavior and Emerging Technologies 3
Correlation?
Fact checked
text explanation
Emotion Explanation Long term
generated acceptance belief change
Creative
explanation
Correlation?
Figure 1: Belief change comparison: fact-based and CC-generated creative explanations.
perspective, we offer the community continuation of the and conspiracy theories [45]. Therefore, the topic is relevant
work by Noorazar through the review of works published from an online belief standpoint. The literature offers a sys-
from 2020 and newer. Urena et al. [37] focus on how trust tematic literature review of definitions and measurements of
propagates in networks from a vantage point of opinions OPP [46]. The finding most relevant to online belief is that
and recommendations. OPP is not an online equivalent of traditional offline politi-
Trust propagation in our context addresses how trust cal participation. It is instead shaped by and contingent
in specific opinion formation agents, such as influencers upon the online platform on which the participation is con-
(INFLs), affects belief formation, a completely different ducted. The work does not elaborate further on the specific
approach. characteristics of platforms that enforce political beliefs that
Opinion formation models are another branch of opin- influence OPP. We hope our survey will provide further
ion dynamics. Mastroeni et al. [38] specifically focus on insight into this topic.
agent-based models, which are centered on their mathemat- Opinion formation is also important in the domain of
ical formulation. Similarly, Abid et al. [39] also focus on the online marketing. Specific to online marketing, product,
mathematical formulation of agent-based models. In con- and service reviews and ratings are driving forces of online
trast, we purposely exclude these works and only include opinion formation. A systematic literature review and
the ones that have some practical validation through either comparative study on how reviews and ratings influence
simulation or dataset experimentation. opinions on buying and usage of the products were pre-
sented in [43]. The work concludes that regular consumer
2.2. User Studies. From a user study perspective, the litera- reviews are more influential in opinion formation than rec-
ture offers a few review works centered on specific online ommendations by professionals and paid experts. It may
information domains, such as health [40, 41], politics [42], be seen as a use-case example of a social influence–based
and online marketing [43, 44]. In the health domain, Wang opinion model for the online marketing domain. Our work
et al. [40] executed a systematic literature review addressing will attempt to find approaches that can be applied across
misinformation spreading online health information. This domains.
study was performed before the COVID-19 pandemic,
which makes it interesting from the standpoint of state-of- 2.3. Methodologies. From the methodology dimension per-
the-art before the event that has dominated health misinfor- spective, the authors of [47] review belief dynamics pro-
mation studies since 2020. While the methodology applied cesses from psychology, sociology, economics, philosophy,
in the work was thorough and the findings around misinfor- biology, computer science, and statistical physics perspec-
mation in online health insightful, the work does not address tives. The work proposes a framework to enable compari-
the vantage point of online belief change. The literature also sons of different belief-capturing methodologies. Even
offers works on COVID-19-related disinformation. Conspir- though individual belief is included as a structural compo-
acy theories are directly related to belief formation and opin- nent of the framework and is briefly discussed, the frame-
ion spread. Different conspiracy theories were born during work limits its modeling to a typical statistical physics
the first year of the COVID-19 pandemic. Tsamakis et al. approach. Our work is aimed at a more holistic view of the
[41] performed a systematic literature review on COVID- existing methodologies and studies the ones best suited to
19-related conspiracy theories. It focused on their preva- capture belief change. In our work, we felt that it is appropri-
lence, determinants, and public health consequences. An ate to combine models and methodologies into a single sec-
interesting result presented by the work, albeit somewhat tion named opinion dynamics, presented in Section 5.3.
predictable, was the higher prevalence of politically moti-
vated COVID-19 conspiracy theories than other determi- 2.4. Polarization. Lastly, for the polarization dimension, a
nants. The work presented studies in the dimensions of notable systematic review links social media to polarization,
demographics, level of income, psychological factors, reli- synthesizing the contingent factors and underlying processes
gion, political orientation, and trust in science. However, it [48]. The work provides three aspects of polarization
marginally addressed online beliefs related to acceptance of conceptualization: the ideological or opinion-based concept,
this class of conspiracy theories, a topic covered by our work. the affective concept of disliking people from outgroups, and
Politics is another online belief-related research hotbed. the social concept of avoiding the company or linkage with
One particular vantage point is online political participation outgroups. This leads to presenting a conceptual framework
(OPP). The level of OPP has been linked to disinformation of social media and polarization. Another study reviews
political polarization from a psychology vantage point [49]. RQ is aimed at identifying whether the literature offers alter-
The work provides a functional conceptualization of polari- native explanation methods beyond the usual fact-checked
zation in an attempt to explain how polarization may occur textual ones. The discovery of alternate explanations may
across partisan fault lines. It provides arguments that polar- modify and expand the high-level experiment design in
ization is most likely to occur in scenarios of belief conflicts Figure 1 into other dimensions of comparison between cre-
in society, such as in politics. Situations of belief conflicts ative explanations and other existing types.
tend to drive the formation of opposed belief groups, which
are prone to polarization. Even though these works provide 3.5. RQ5—How Can the Identified Belief-Changing Approaches
a rich link between online beliefs and polarization, neither Be Generalized to Multiple Belief Domains? This RQ will seek
studies the effect that polarization may have on constraining insights into whether any of the existing belief-changing
belief change. Our findings from that vantage point are pre- approaches have the potential to be generalized into a frame-
sented in Section 5.2. work that can cover multiple belief domains. We will high-
light their strengths and weaknesses from a generalization
3. Research Questions potential standpoint. We will conclude the analysis by pro-
viding recommendations toward the generalization goal.
Figure 1 illustrates the conceptual idea for the future
research setup to validate our main hypothesis. However, 4. Methodology
several questions remain unanswered regarding the detailed
implementation of this concept. The research questions pre- This section presents the methodology used to create the
sented in this section were designed to help answer some of final corpus of work reviewed. It explains the inclusion/
these questions. exclusion criteria, the methodology applied, and the taxon-
omy used to classify the reviewed works.
3.1. RQ1—What Are the Main Drivers of Online Belief
Change? Understanding the primary motivators that lead 4.1. Inclusion and Exclusion Criteria. This work includes
online users to change their preconceived beliefs is pivotal peer-reviewed journal articles and conference papers of
for designing personalized explanations that efficiently edu- more than four pages in length, published in 2020 and later
cate users in the event of fake news beliefs. This understand- found in the following databases: Scopus, ACM, IEEE
ing is also central to the design of experiments to validate the Xplore, and Web of Science. Nonpeer-reviewed articles and
main hypothesis of this work. book chapter papers not centered on online belief and opin-
3.2. RQ2—How Are Current Opinion Models Being Used to ion change are excluded from the review. Opinion dynamics
Capture Belief Changes? Opinion dynamics covers a wide theory works are included when validated via dataset or
range of social science phenomena, such as the appearance simulation-based experiments. Purely theoretical papers are
of fads, consensus building, collective decision-making, excluded. These papers are deemed too far removed from
rumor spreading, extremist expansion, and even cult propa- the goal of hypothesis verification as they would still need
gation [50]. This RQ constrains the analysis of the models to to be validated through experimentation. One of this work’s
focus on online belief change. RQ2 intends to understand if goals is to understand potential psychological and social
there are specific models offered by the literature that can be forces that may constrain acceptance of fact-based explana-
applied to collected experimental data to facilitate the iden- tions. Therefore, polarization papers focusing solely on algo-
tification of belief change by the participants. rithm bias effects on polarization are excluded. Table 1
summarizes the inclusion and exclusion criteria.
3.3. RQ3—What Role Does Polarization Play in Changing An initial search according to the following keywords
Online Users’ Preconceived Beliefs? This RQ explores the (online OR “social network”) AND (“opinion change” OR
direct effect of information bias and polarization in changing “belief change” OR “change in belief” OR “opinion forma-
online users’ beliefs. This is important since the chief objec- tion”) filtering peer-reviewed papers published in 2020 or
tive of fake news systems is to align news consumers with sooner returned 372 hits. A visualization created with the
factual news. For that to happen, people with preconceived VOSviewer software application [51] was performed to iden-
beliefs in fake news shall be shown that their beliefs are tify potential clusters. Figure 2 presents the initial visualiza-
not based on facts and ought to change. Understanding tion results. Four high-level clusters were identified: user
not only if polarization is an important force potentially pre- studies (green), opinion dynamics related to opinion forma-
venting belief change but also if there are documented tion (red), belief related to user intervention (purple), and
approaches to best deal with this driver may provide insight public opinion (blue). Furthermore, the user studies cluster
into the explanation content and presentation that carry the revealed the specific domains of COVID-19, climate change,
best odds of success. Furthermore, it may add a dimension education, and politics.
to the experiment as we can compare the polarization effect There are some noticeable correlations between the visu-
on purely factual and personalized explanations. alization of Figure 2 and the taxonomy presented in Section
4.2. We originally named the red cluster “opinion forma-
3.4. RQ4—What Alternative Approaches to Offering Fact- tion” due to the highest prevalence of the word “formation”
Checked Explanations Have Been Pursued by the Literature in the visualization. However, the reading of the works
in an Attempt to Change Preconceived Online Beliefs? This revealed that the highest prevalence of the word was not
Table 1: Inclusion and exclusion criteria.
Inclusion Exclusion
Peer-reviewed journal articles and conference papers longer
Workshops, book chapters, surveys, and nonpeer-reviewed papers
than four pages in length
2020 and newer 2019 and older
Opinion dynamics modeling belief change validated by Theoretical models only. User studies and simulated models not
experimental data affecting belief change or formation
Polarization affecting beliefs Algorithm bias effect on polarization
Domains User intervention

User studies Public opinion
Opinion dynamics
Figure 2: Opinion and belief change 2020–2023.
necessarily due to the cluster being about opinion formation The blue cluster shows the term “public opinion” as
but because of its relationship with the “opinion dynamics” highly prevalent because of its orthogonality with all three
subject. This is why the word “relationship” is also highly taxonomy dimensions. In this context, “public opinion,”
prevalent in the obtained visualization. “public opinion formation,” and “network public opinion”
Terms with lower prevalence than opinion dynamics but were combined in a single cluster: group opinion (GRPO).
with significance, such as “opinion formation process” and The high prevalence of “impact,” “topic,” and “factor” is
“opinion formation model,” are, in fact, opinion dynamics somewhat synonymous in our context. The green cluster
implementations either through simulation or dataset exper- groups the user study papers. The identified user study
imentation. This is the reason that the opinion dynamics domains in the visualization, COVID-19, student, citizen,
dimension is subdivided into simulation and dataset experi- and climate change became taxonomy clusters. The terms
ment clusters. Other terms such as “structure,” “opinion evo- “support” and “exposure” were related to the sex and homo-
lution,” and “network topology” represent applications of sexuality theme, which was turned into a taxonomy cluster.
opinion dynamics techniques. The techniques can either The visualization motivated adding other keyword
focus on network structure (NETS) or the dynamics of opin- searches centered on the following topics: belief and opinion
ion evolution in different contexts. These contexts are, changing user studies, belief formation models and method-
namely, situations of crisis (CRIS) or traumatic events, the ologies, opinion formation models and methodologies, and
influence of stubborn or strong opinioned neighbors and polarization. The set of keywords for each search is illus-
INFLs, analysis of confirmation bias or homophily (HOMY), trated in Figure 3.
analysis of sociological or psychological forces in opinion Applying these criteria to the corpus yielded a total of 91
dynamics, and the study of group or public opinion forma- papers that were reviewed. These papers received a com-
tion. All of these contexts became orthogonal clusters of the bined 782 citations at the time of this writing. Sixty-six of
taxonomy as they are relevant to all dimensions of the review. these works were published in journals, and the remainder
Participant
And Dataset
OR
Experiment not
And
Study
Opinion change
OR
Belief change
OR Model
Online Change in belief OR
OR OR Method
And And
Social network Opinion formation OR
OR Methodology
Belief formation
OR
Opinion creation Bias
OR
Polarization
OR
Echochamber
Figure 3: Keyword combinations.
in conferences. Fifty-one percent was published in Q1 jour- 4.2.6. GRPO. It focus on belief change of groups and public
nals and 10% in Q2 journals. Seven percent was published in opinion.
A conferences and 9% in B conferences.
4.2.7. SNSBs (SNS Biases). It focuses on SNSBs, which
4.2. Taxonomy. The application of the methodology pre- include filter bubbles [144] and other bias-inducing algo-
sented in Section 4 drove the organization of this review into rithms used by social networking sites.
three dimensions: domain specific, opinion models, and
polarization. 4.2.8. NETS. It focuses on the influence that neighbor agents
Table 2 shows the taxonomy classification and corre- may have on belief change for groups within the same
sponding works assigned to clusters. Some papers appear network.
in more than one cluster. The description of each cluster is
presented below. 4.2.9. CRIS. It includes CRISs that did not lead to TRMA.
4.2.1. TRMA (Trauma). It includes papers addressing belief 4.2.10. STROs (Strong Opinions). It includes papers that
change as a result of traumatic events. TRMA has been address agents with STROs about a subject, stubborn, and
defined as “the experience of a vital discrepancy between zealot agents.
threatening factors in a situation and individual coping abil-
ities” [138]. TRMA can be objective and subjective [139]. 5. Findings
Objective traumatic TRMA directly leads to post-traumatic
This section presents the survey findings in the context of its
stress disorder (PTSD). Subjective traumatic events may or
taxonomy.
may not.
5.1. Domain-Specific Dimension. As shown in Table 2, the
4.2.2. INFL. It includes the effect of INFL [140] agents in
domain-specific dimension of the review was split into five
belief change.
classes: COVID-19, climate change, education, politics and
4.2.3. HOMY. HOMY is attributed to people’s natural ten- policy, and other. This section presents the findings for this
dency to associate with people similar to themselves. Studies dimension. Trust can be defined as “a psychological state
have documented that even infants as young as 6 months of comprising the intention to accept vulnerability based upon
age already show HOMY [141]. positive expectations of the intentions or behavior of
another” [145]. Trust is, therefore, a psychological construct,
4.2.4. BCHB (Biaschamber). We dubbed BCHB as a combi- and it was one of the central drivers of opinion formation
nation of echo chamber and confirmation bias. An echo [79, 80] and belief change for the works reviewed in this
chamber is defined as the formation of like-minded online dimension [52]. Trust in celebrities, namely, parasocial rela-
users reinforcing a narrative [142]. Confirmation bias is tionships [65], and social network INFLs was exploration
defined as the seeking to interpret evidence in ways that topics. It was seen that INFLs can significantly affect people’s
are partial to existing beliefs [143]. opinions on different issues [52, 54] and that trust in the
information source is an essential driver of belief change.
4.2.5. PSOC (Psychosocial). Even though HOMY and BCHB Trust in government and officials was correlated with the
are PSOC phenomena, this cluster includes other sociologi- consistency of the public messaging, affecting online belief
cal and psychological constructs. change [70].
Table 2: Taxonomy of included papers in review.
TRMA INFL HOMY BCHB PSOC GRPO SNSB NETS CRIS STRO
Domain specific
COVID-19 [52–54] [55] [56] [54–56] [55]
Climate change [57] [58] [59]
Education [60] [61, 62] [60] [60] [63]
[65, 68,
Politics and policy [64] [65, 66] [67] [70] [66] [69] [64]
69]
Sex and
[71, 72]
homosexuality
[74,
Other [73] [18, 74] [75] [79] [75] [79, 80]
76–78]
Opinion dynamics
Simulation [81–83] [84] [85] [83, 86] [83, 85–93] [94] [95–97]
Dataset [101, [86, 98, 100,
[98–100] [102–106] [86, 100] [107]
experiments 102] 103, 107, 108]
Polarization
Theoretical studies [102] [102] [102] [109] [110]
[111, [114, [111, 123, [125,
Models [113] [16, 116] [117, 118] [119–122]
112] 115] 124] 126]
[129, [127,
User studies [127] [128] [131–133] [134, 135] [136]
130] 137]
Other psychological constructs were addressed in this relationship between core belief challenge and PTG. Still,
dimension. Normative influence [68] is popularly known in psychology, the authors of [76] attempted to understand
as herd mentality. Normative influence originates with the whether specific personalities are more prone to be per-
basic human desire to not stand apart from a group, i.e., suaded into changing their beliefs, but their results were
the desire for social acceptance, which varies according to inconclusive.
one’s perceived risk of social rejection [146]. Informational It was observed that the polarization of subjects seems to
influence is defined as the use of group knowledge as a deter- be cross-correlated. Political affiliation was shown to be a
minant of correct beliefs [147]. In [68], the authors showed polarization topic to be at the center of cross-correlation
that factual information was not the primary driver of voting with topics such as environmentalism and climate ideology,
behavior. Normative influence may be powerful enough to COVID-19 response, policy preferences, immigration and
influence belief change toward conformity to group deci- patriotism, and even beliefs in biological attribution to
sions, even in secret voting. HOMY is another. In [67], the homosexuality [71], which is directly related to support for
authors showed that it is possible to build a profile of former homosexual rights. The results obtained in [54] revealed a
US President Donald Trump’s supporters using HOMY as connection between political viewpoints and misinformation
one of the most predictive signals for the model. Political sci- regarding hydroxychloroquine (HCQ) in treating COVID-
entist Elizabeth Noelle–Neumann proposed a theory dubbed 19 despite not being supported by scientific evidence. The
the spiral of silence [148], highlighting individuals’ unwill- author in [69] showed how political orientation is critical
ingness to publicly express an opinion when they feel that in shaping how public crises are interpreted and how belief
it may be against the majority opinion. This was verified changes about them. The authors of [58] show an associa-
through the opinion of social media users on LGBT accep- tion between left/right political ideology and environmental-
tance in Nigeria [72, 79]. Bandwagon effects refer to individ- ist/skeptic climate ideology, respectively.
uals’ tendencies to conform to predecessors’ decisions [149]. The connectivity between the traumatic public events
In an online scenario, users are likely to rely on and gravitate and the arousal of emotional processes was demonstrated.
toward more popular opinions as a form of mental shortcut. Examples are the historical and institutional racism added
The authors of [74] showed strong evidence of further polar- to historical TRMAs such as the Tuskegee syphilis study
ization of preconceived beliefs away from the expert’s pre- [153] and the unethical and nonconsensual use of cancer
sented opinion. TRMA is another psychological event that cells from Henrietta Lacks [154] providing context for
can drive core belief change [150], with results showing that understanding vaccine hesitancy among Black individuals
people who underwent intense TRMA feel that they changed and their distrust of healthcare professionals and researchers
their beliefs toward humanity. Post-traumatic growth (PTG) [52]. A CRIS also evokes emotional responses that lead to
is defined as how individuals can experience positive psy- polarization. This was demonstrated in [59] for the climate
chological change after a traumatic event [151]. The authors change topic and the topic of public opinion about police
of [73] showed deliberate rumination [152] to mediate the funding [64] at around the time of the murder of George
Floyd at the hands of law enforcement in the United States theoretical studies, models, and user studies. This section
[155]. It was also seen that the level of a person’s stubborn- presents the findings for this dimension.
ness was shown to be inversely correlated to the probability PSOC polarization driving forces were identified,
of belief change [55]. The hypothesized correlation between namely, normative influence [116, 157], spiral of silence
emotional arousal and polarization was confirmed. [148], confirmation bias, backfire effect, parasocial relation-
Furthermore, a strong and direct stance stating the con- ships, and HOMY. Confirmation bias influences polariza-
tent is fake invariably leads to conflict, aligning with the tion as the intensity of preconceived beliefs is sometimes
finding that presenting factual explanations defending an the controlling aspect of belief change [130]. Arguably,
opposite position can sometimes backfire and further people seek communities with higher chances to confirm
entrench polarized people in their preconceived beliefs their beliefs [136]. The results obtained in [16] showed
[16]. Stubbornness can lead to the entrenchment of beliefs. confirmation bias in combination with the backfire effect
This can be seen even in less polarized topics, such as pri- to be strong drivers of polarization. The authors in [133]
mary school teachers’ beliefs about teaching computer sci- showed evidence of polarization development in another
ence [63]. The results of this work showed that younger, combination of PSOC constructs: parasocial relationships
less experienced teachers showed no signs of belief persever- and HOMY. It was shown that people became further
ance. Conversely, older, more experienced teachers demon- entrenched in their preconceived beliefs in the case of a
strated higher levels of belief perseverance, even when they contradicting opinion from a subject matter expert celeb-
indicated positive reactions toward the received computer rity. In [131], the authors showed that feelings of resent-
science training. ment were the most significant predictor of the Black
Multiple studies provided evidence of the efficacy of Lives Matter movement’s support. Low-resentment individ-
explanations that nudge people into a state of reflection uals who expressed themselves on social media more fre-
about their preconceived beliefs [18]. Ruffin et al. argue that quently were less supportive.
attempting to explain how fake photographs were manipu- Some papers demonstrated how some fragmented net-
lated offers better results if done cautiously [18]. The works self-organize into multiple echo chambers of consen-
authors of [62] showed this to be also valid in the context sus and that consensus is a precondition for the emergence
of belief change related to the nature of intelligence. Their of polarization [109, 113, 114, 126, 128]. The authors of
results showed that rather than convincing people that [115] looked even further into the correlation between echo
intelligence is malleable, gentle mindset interventions may chambers and polarization. The authors argued that their
be the most important activity for helping them reflect on results validated the idea that echo chambers create a stable
intelligence’s malleability. This nudging may happen with environment of confirmation bias and can even actively
the help of an emotion-evoking explanation, for example. alienate some group members from outside contradicting
Emotion responses were correlated to low knowledge in the information sources [158]. Similar results were obtained in
process of a layperson acceptance and resultant opinion for- [117, 129]. Another relevant finding was that if the same
mation related to climate engineering approaches [57] and argument is presented by two people, one from their com-
the driving of belief change of teachers under online and munity and the other from another network, the likelihood
blended delivery methods [61]. The results showed that of acceptance of the former is notably higher. This suggests
increasing knowledge about the topic in both cases drove that one possible way to reduce polarization may be to
belief change. This validates the concept that if knowledge change beliefs from within. Focusing on changing the beliefs
is low regarding a given topic, emotional responses are used of key members, such as INFLs, of a polarized group may
as indicators for attitudes toward or against a stimulus trigger a snowball effect in the beliefs of all members of the
[156]. It also reinforces the need to balance an emotion- given community. The results in [123] suggest that this
evoking explanation with facts to drive an increase in sub- may be the case as they showed that most individuals from
ject knowledge. a network over time switch to opposite sentiments about
Explanation personalization was also addressed in this the preconceived belief. The results obtained in [111] suggest
dimension. The authors of [77] showed that a personalized that another possible way to revert polarization is to shield
online algorithm–based intervention can change beliefs that the members from their corresponding echo chambers,
may lead to inappropriate antibiotic demand by patients. allowing them to access the ideas of members outside these
Conversely, results obtained in [75] show that personaliza- chambers freely.
tion enhances user experience, but the so-called “filter bub- Evidence also acknowledges that user adherence to mis-
bles” favor the emergence of opinion polarization and information may sometimes be shifted away from accuracy
radicalization through confirmation bias. One final notewor- and toward other goals. In [102], the authors concluded that
thy comment is about an interesting approach using senti- providing subtle accuracy nudges is a promising approach to
ment analysis (SENTANL) pre- and postevent to capture improving the quality of shared news. The correlation
belief change [56]. This methodology is promising and between SNSB and polarization was analyzed and verified
should be investigated further as a potential approach to val- [119]. Arguably, there is also a correlation between SNSB
idate this work’s central hypothesis. and individual PSOC constructs. Correlations between
polarization due to the spiral of silence [121] and filter bub-
5.2. Polarization Dimension. As shown in Table 2, the bles [122, 159] were demonstrated when people are influ-
review’s polarization dimension was split into three classes: enced by strong SNSB. Another study looked at SNSB
through the lens of how people change their opinions when Some studies highlighted the importance of a solid fac-
exposed to viral content [120]. The results showed that tual foundation to balance emotional arousal that nudges
polarization barely increased after a regular marketing cam- people onto reflective states for a higher probability of
paign and significantly increased upon the spread of polar- changing polarized beliefs. Emotion was confirmed to be
ized content. an important component in this nudging, especially when
The cross-correlation between polarized subjects also balanced with other cognitive functions. Emotion was
becomes evident after the review of this domain. It seems shown to be correlated with the higher interest people
that political ideology is the central topic of polarization, showed in resharing audio messages than purely text mes-
and it can become cross-correlated with other polarization- sages on social networks [102]. We hypothesize that audio
prone subjects such as minority equality [137], patriotism, messages have the potential to carry more emotional con-
welfare policy [135], and the response to health crises tent than textual messages, driving people to have more
[134]. The authors of [132] demonstrated that this cross- interest in resharing them. The results in [85] showed that
correlation directly correlates with emotion. They concluded an online post combining affective and cognitive content
that a psychological factor that impedes climate change increases people’s willingness to share the message. Con-
beliefs is not related to climate but is mainly motivated by versely, effectively weak and mostly cognitive content was
the feelings of dislike one political group feels toward the shared the least. The nudging also needs to be founded on
opposing group. CRIS events were also connected to the facts. The authors in [86] showed how removing facts from
emergence of polarization [127, 137]. a post alienates people, and this alienation drives the
emergency of nonfactual subtopics. The formed new sub-
5.3. Opinion Dynamics Dimension. This section presents the groups can lead to a phenomenon known as information
findings for the opinion dynamics dimension. It includes the gerrymandering [160], where STRO individuals can keep
research works split into two classes, simulation and dataset negatively held opinions alive, even if nonfactual, as dem-
experiment, as defined in Table 2. onstrated in [96].
The study of the effects of PSOC constructs was also Information alienation was highly correlated with the
present. HOMY is an important one. The results in [104] emergence of polarized subnetworks. Therefore, it is impor-
showed the effects of HOMY in the formation of echo cham- tant to share information about a given topic of interest to
bers. It also demonstrated a moderate to high resemblance of public opinion as early as possible, especially during a CRIS
the echo-chamber phenomenon for network topologies of [94]. However, this needs to be done carefully to avoid a sce-
abortion, capitalism, and feminism. This aligns with trends nario of inconsistent messaging in case the results need to be
from other dimensions, suggesting cross-correlation between reviewed later. The information revision may cause a back-
polarization topics. In [106], the authors show a context of fire effect as [70] has provided evidence that inconsistent
evolving HOMY in political social network interactions. messaging reduces the effectiveness of explanations targeted
The results in [83] showed how HOMY and the spiral of at changing group beliefs. The constant changing of messag-
silence drive people to form online social groups. The band- ing was shown to generate a breach of trust by the public
wagon effect, or herd mentality, influences consensus forma- concerning the source of the message.
tion, as verified in [149]. The author in [101] showed a Research from this dimension found evidence that large
tendency for moderate online users to move toward the aver- networks with a diversion of opinions evolve into several
age opinion of their online friends. The authors in [98] smaller networks where consensus is reached and then
showed that the bandwagon effect has a stronger driving polarization develops [87, 89]. However, Mansouri and
force than INFLs and that the reach of consensus will be Taghiyareh [82] show that when influential leaders exist in
magnified in a scenario of bandwagon effect. However, this a social network, segregation has less impact on opinion for-
does not happen in highly segregated opinion networks mation than the effect created by INFLs. This shows how
[93]. This is an important finding as it suggests that polariza- INFLs are key drivers of belief change in opinion networks
tion can be avoided if education on fake news posts happens [97], including public opinion formation [92]. This effect
at the initial stages of a social network before its consequent was also verified when mass media played the role of INFLs
evolution to consensus. It was demonstrated in [81] that it [84]. It was shown that even a small percentage of INFL-type
is more difficult for someone to reach a consensus with a per- agents motivated to manipulate opinion toward a specific
son who belongs to a group with a higher proportion of low- goal could shape the majority opinion [100]. Similar results
educated people than with a higher proportion of high- were shown in [99].
educated people. Another data point that shows the impor- Being the intermediate step between opinion segrega-
tance of educating online users on fake news posts. PSOC tion and polarization, consensus needs to be understood.
constructs are part of what forms a person’s individuality. The results in [95] reveal that consensus in a multitopic
Individuality is also important regarding how personal expe- network can be achieved if the number of stubborn agents
riences help shape GRPOs. The authors in [103] showed how around the subjects is small. Lastly, natural language pro-
GRPO results from the community’s combined individual cessing (NLP) SENTANL on social network posts to iden-
experiences. The authors argue that the so-called expert tify belief change was also present in this domain [108].
agents, or agents that bring strong individual experiences This seems to be the preferred technique for identifying
aligned with subjects of interest to the group, are highly influ- belief change by online users and is used across applica-
ential to group beliefs. tion domains.
6. Discussion Table 3 shows that four cross-dimensional trends were

addressed by reviewed works from all three dimensions:
Section 5 presented the findings for the three dimensions PSOC, INFL, CRSPOL, and EMLNDG. Of the four, PSOC
included in this review. It presented the trends found in each had the most representation, with 32 papers. Several PSOC
dimension separately. As the chief focus of this work is a constructs were present in these 32 works. Confirmation bias
holistic review of online belief change, we consider the was shown to have a positive correlation with the openness
trends that appeared in more than one of the investigated personality trait and a negative correlation with neuroticism
dimensions to be the most relevant to belief change. This [161]. These two personality traits are part of the big five
section summarizes these trends, and Table 3 shows the personality model [162]. The entrenchment of beliefs is a
number of reviewed papers that addressed each of them in complex construct that may have several root drivers; how-
their corresponding dimensions. A given reviewed paper ever, the personality or character of the believer has been
may have included more than a single trend. Conversely, identified as an important factor [163]. Attitudinal HOMY
some reviewed papers may have addressed a trend that did refers to personality and attitude similarities between indi-
not appear in multiple dimensions. The following acronyms viduals [164]. In the context of celebrities, the more a person
define each cross-dimensional trend. identifies similarities between a celebrity’s attitudes and
overall personality and their own, the more this individual
• SUBCNS (subconsensus): Several works reviewed will research about that celebrity [165], leading to parasocial
showed that an initially segregated network naturally relationships. Previous studies have demonstrated HOMY as
self-organizes into multiple echo chamber subnet- one of the predictors of parasocial interactions [166]. There
works of consensus and then further evolves into cor- is currently no consensus on the meaning of HOMY beyond
responding polarization groups. the broad definition stating that like-minded people tend to
form communities. However, some psychology researchers
• FACBAL (factual balance): Multiple works showed
argue that one’s personality defines one’s HOMY nature
that a combination of knowledge increases positive
[167]. Previous research correlated the spiral of silence with
emotion, gently nudging online users into a reflective
the cultural behaviors of individualism and collectivism
state. It became clear the importance of explanations
[168]. Emotion research theory considers culture to be one
to balance facts and emotional evocation for increased
of the three driving influences of how people perceive and
odds of success in belief change.
act on emotions [169]. Therefore, it is plausible, albeit still
• INFL: INFL in SNS was confirmed to be a high driving not confirmed, that cultural differences may offer a correla-
force of belief change. tion between the spiral of silence and emotion. Research is
evolving toward an irrefutable connection between these
• SENTANL: NLP SENTANL before and after a specific psychological constructs and individual personalities. How-
event seems to be the preferred technique for identify- ever, the correlations should not be ignored.
ing belief changes by online users. The second ubiquitous trend with the most representa-
• PSOC: This cross-dimensional trend includes the tion in the review is INFL, with a total 19. Trust in the infor-
PSOC constructs linked to belief change. Psychosocio- mation source is the foundation for the INFL’s driving force
logical tendencies such as filter bubbles, spiral of in SNSs, its development ranging from parasocial relation-
silence, confirmation bias, backfire effect, parasocial ships in the case of celebrity INFLs through confirmation
relationships, and HOMY facilitate the evolution of biases and other intrinsic individual tendencies. A possible
social network groups into a scenario of polarization. conclusion is that trust in INFLs has its roots in psycholog-
Trust was also shown to be a psychological construct, ical constructs. Therefore, the INFL trend may be considered
arguably one of the strongest drivers of belief change a corollary of the PSOC trend in our context. Psychology
in this group. research has also shown a direct relationship between psy-
chology, personality, and emotions [170–172]. We argue
• CRSPOL (cross-polarization): All three dimensions that this close relationship between personality and emotion
showed that subjects’ polarization is cross-correlated may be why EMLNDG also appears in all three dimensions
in different subjects, with political ideals being the cen- of this work, addressed by 14 papers.
tral polarization topic. CRSPOL is the last cross-dimensional trend that covers
• ERLPUB (early publication): Several works showed the all three domains, albeit with a total number of papers much
importance of publishing results debunking false lower than the other three trends. INFLs have a strong effect
claims as early as possible to counteract public disin- in driving polarization, which emerges in good part due to
formation. However, this needs to be done carefully psychological or emotional reasons or a combination of
to avoid a scenario of inconsistent messaging in case both. It is expected that a topic chiefly influenced by the
the results need to be reviewed later. three main drivers of belief change would also be present
in all three review domains.
• EMLNDG (emotional nudging): EMLNDG is referred There is a strong relationship between emotion and sen-
to herein as the alternative to a factual explanation that timent, as sentiment can be construed as a thought, an opin-
balances facts with emotional evocation to nudge the ion held by the person based on a feeling. In general terms,
online user to a reflective state. sentiment is the effect of emotion [173]. Since emotion plays
Table 3: Reviewed papers addressing each cross-dimensional trend.
SUBCNS FACBAL INFL SENTANL PSOC CRSPOL ERLPUB EMLNDG Total

Domain specific 0 3 8 2 16 5 1 8 43
Polarization 5 0 4 0 11 3 0 4 27
Opinion dynamics 4 2 7 1 5 1 1 2 23
Total 9 5 19 3 32 9 2 14 93
such a pivotal role in belief change, it becomes natural that CRSPOL, as well as the other identified cross-dimensional
SENTANL emerged as a cross-domain trend and the pre- trends, was corollary to the two main ones. This numerical
ferred method for evaluating online belief change. analysis indicates that PSOC constructs and emotional
Perhaps contrary to intuition, purely factual explana- arousal are arguably the two main drivers of online belief
tions are not the most efficient in changing online forged change.
beliefs. EMLNDG’s importance to belief change drives the Psychology research has also shown a strong correlation
corollary cross-dimensional trend of FACBAL. FACBAL between individuality, personality, and emotions. Tellegen
focuses on balancing facts and emotional arousal in explana- [174] has proposed that even though environmental changes
tions. The ERLPUB cross-dimensional trend is a natural may influence affective responses, a full appreciation of indi-
consequence of the potential breach of trust between public vidual differences in emotional response could only be per-
opinion and officials who publish erroneous early communi- formed if personalities and how they influence affect are
cations and are forced to review the message later. The evo- considered. The authors of [175] performed a user study
lution of belief change from a fragmented network through and concluded that personality is an essential determinant
the formation of subnetworks of consensus that eventually of an individual’s emotional response. Moreover, in [176],
lead to polarization is the central topic of the SUBCNS a user study shows that individuals who present with high
cross-dimensional trend. negative affectivity are generally more introspective, a per-
In summary, we showed that the cross-dimensional sonality trait, and are more likely to experience discomfort
trends present in all three dimensions of our work are driven at all times, even in the absence of stress. This shows that
either by PSOC constructs, emotion, or a combination of the individuals perceive emotions differently.
two. We showed that the other cross-dimensional trends Personality and individuality have been treated as syno-
presented have roots in these two drivers. Therefore, we nyms by various English-language dictionaries. Personality
argue that PSOC constructs and emotions are the two main has been defined as “the incarnation of individuality”
drivers of online belief change. The following section will [177]. The strong correlation between personality and emo-
present answers to each of the proposed research questions. tion suggests them to be individual characteristics. There-
fore, we argue that individuality is the most critical driver
6.1. Research Questions Answered. This section provides of online belief change, materialized through psychological
answers to research questions that emerged from the traits and emotions. This result partially validates this work’s
reviewed works. central hypothesis that personalized explanations are more
efficient in reducing fake news spread.
6.1.1. RQ1—What Are the Main Drivers of Online Belief
Change? The discussion in Section 6 presented the cross- 6.1.2. RQ2—How Are Current Opinion Models Being Used
dimensional trends, and Table 3 shows the breakdown of to Capture Belief Changes? The current opinion models
the number of papers that addressed each one of the trends. used in the reviewed works that either performed simula-
A numerical analysis of Table 3 indicates PSOC constructs tions or used real datasets to perform experiences yielded
to be the top ubiquitous trend, addressed by 34.4% of all important conclusions in capturing belief change. Interest-
papers. INFLs were the second most addressed trend by ingly, the opinion dynamics dimension works contributed
20.4% of all papers. We did argue, however, that trust is a to all eight cross-dimensional trends in Table 3. It is impor-
psychological construct, and it is at the center of the INFL tant to note how these works help to model the evolution of
drive for belief change. This argument suggests that both opinion dynamics, starting from regular social networks
trends can be combined, leading to over half, or 54.4%, of into multiple subnetworks of consensus and ultimately into
all reviewed papers to have focused on PSOC constructs polarization.
for belief change. Within the context of each dimension, The most popular approach for capturing belief change
combining the two trends resulted in 55.8% of the domain- is using NLP SENTANL models in social media posts
specific works, 55.5% of the polarization works, and 52.1% [178]. The overarching concept is to perform a sentiment
of the opinion dynamics works. This shows an equivalent temporal analysis [179] of posts before and after an event
balance of relevance within each of the domains. EMLNDG with the potential to drive belief change to verify sentiment
accounts for a total of 15.1% of all reviewed works. This change over a specific subject. In the context of our research,
trend is the distant next highest trend, but it is much more a given fake claim is the subject, and the provided explana-
prevalent than CRSPOL, the last ubiquitous trend, which tion is the event of interest. A secondary approach that has
appears in just 9.6% of all papers. We argued, however, that been gaining momentum is the temporal analysis of patterns
of emotions associated with social media posts [180]. This erature does not seem to offer many alternatives to fact-
approach focuses on performing a lexicon-based analysis checked explanations. Some works evaluate whether fact-
measuring valence, arousal, and dominance of social media based explanations are efficient in changing beliefs; however,
posts using the VAD Lexicon [181]. The works reviewed also they have not attempted to apply alternate methods. Even
showed how INFLs help drive the evolution of opinion in though we could not find a direct answer to this RQ, it moti-
social networks. It was demonstrated that INFLs are very vated the emergence of an interesting conclusion. The eval-
important in shaping the beliefs of the subnetworks of con- uation of efficiency in changing online users’ beliefs by a
sensus. Moreover, it was also shown how INFLs who manip- purely factual explanation was performed in different
ulate information for some personal gain seem to have an domains: images, audio, and text messages. The studies
even greater driving force in the creation of polarized net- come to the conclusion that factual explanations are ineffi-
works. This is critical information for creating explanations cient in changing preconceived beliefs in all presentation
that can efficiently change online users’ preconceived beliefs. domains and that a gentler approach should be investigated.
Since INFLs are key to forging opinions, they can potentially
be critical agents to start a snowball effect of belief change 6.1.5. RQ5—How Can the Identified Belief-Changing
toward a well-balanced factual explanation debunking fake Approaches Be Generalized to Multiple Belief Domains? The
news. Therefore, the models that identify social network answer to RQ4 showed the inefficiency of purely factual
INFLs are also important to belief change. More specifically, explanations across all presentation domains. The same con-
the models that find context-based INFLs that polarize these clusion was reached by works looking at different subject
subnetworks [182] can be used to help focus the application domains. This aligns with our hypothesis that explanation
of the explanation on these INFLs, followed by the applica- personalization can be the nudge to drive users toward that
tion of temporal SENTANL models to verify whether the deliberate thinking-reasoning process. The original goal of
explanation was effective in changing their beliefs. this RQ was to find out if the identified belief-changing
approaches can be generally applied to multiple fake news
6.1.3. RQ3—What Role Does Polarization Play in Changing domains. As stated, the studies did not specifically reveal
Online Users’ Preconceived Beliefs? The highest driving forces alternate methodologies; however, the fact that studies in
of belief change were also central to creating or expanding many different domains recommended the concept of nudg-
polarized online scenarios. Psychological traits and INFL ing people into reflective states indicates that an approach of
trends account for 55.5% of all reviewed works in this balanced explanations as previously stated may be efficient
dimension. These papers confirmed that HOMY, confirma- across domains.
tion bias, and trust in the information source are especially
influential in belief change within polarized networks. More- 6.2. Grand Challenges and Future Work. This section pre-
over, the SUBCNS trend, covered by 18.5% of the reviewed sents the grand challenges that emerged from the identifica-
works in this dimension, presented an important characteris- tion of the cross-dimension trends.
tic of belief change toward polarization. It was concluded that
a condition of polarization can be very easily created in SNSs. 6.2.1. Psychology Research Intersection. This work argues
The evolutionary process of polarized network formation that individuality is the chief driver of online belief change
starts in opinion-segregated networks, advances to multiple through its exteriorization as personality traits and individ-
subnetworks of consensus, and settles in many polarized net- ual emotional responses. Several psychological constructs
works attracted by and formed around INFL agents. It was were presented as being of influence in opinion formation.
seen that once polarization is established, entrenchment It would be important to advance online belief change and
and backfire effects are typical psychoemotional responses fake news explanation research to have a more mature foun-
by polarized individuals to factual explanations that contra- dation of psychology research showing solid relationships
dict their preconceived beliefs. These individuals become between given personality types, emotional responses, and
somewhat immune to fact-based correcting information and, the psychological constructs identified as important for
therefore, much more resistant to belief change. Two impor- online belief change. Even though the field shows meaning-
tant conclusions can be drawn. Firstly, it shows the importance ful correlations that should not be dismissed, research in this
of balancing factual explanations with enough emotional con- area is still evolving.
tent to gently nudge these individuals into reflective states to As an example, we have shown that explanations that are
work around entrenchment situations. Secondly, changing an balanced between factual and emotional content carry a
INFL’s belief from within a polarized community may trigger higher potential to avoid preconceived beliefs entrenchment.
a belief change snowball effect. This motivates the hypothesis However, it is essential to consider that the same content
that a belief-changing approach could combine the two con- may elicit different emotions in different explanation recipi-
cepts: adding emotional content to the explanation to reduce ents. It is plausible that a given explanation that is expected
entrenchment, focus, and personalize these explanations on to elicit a positive emotion to nudge the recipient into a
the INFLs of a polarized network. reflective state may backfire and generate a counter-
productive one that may drive entrenchment. A deeper
6.1.4. RQ4—What Alternative Approaches to Offering Fact- understanding of how people of different personalities react
Checked Explanations Have Been Pursued by the Literature to emotions could provide more qualified information to be
in an Attempt to Change Preconceived Online Beliefs? The lit- used in a deeper personalization of the content to maximize
positive emotions. Without it, the results may be negatively checked text. Creative explanations can be defined as involv-
biased if an imbalance of personalities is involved in the ing some level of the creative process to generate an output
method evaluation. With these limitations in mind, future that can be considered creative. Creativity is immediately
research investigating this hypothesis should include exten- connected to art. Art arguably sparks experiences that simul-
sive demographic and cognitive preference data to charac- taneously engage many aspects of an individual’s mental life,
terize the study participants as much as possible. This including emotions [186]. Art in this context can be
approach may offer opportunities for cross-correlation of expanded to its multiple domains, such as poetry, music,
emotional responses with individual attributes that may painting, and others. Humor has been shown to include pat-
shed light on potentially unexpected or contrary results. terns of intercorrelations with several measures of creativity
[187] and is a vehicle for emotional arousal [188]. In high-
6.2.2. INFL Detection. This work suggested that targeting level terms, as long as the explanation is anchored in facts
well-balanced explanations on INFL agents may cause a pos- to avoid the risk of misinforming the reader, any explanation
itive snowball effect to break polarization. This requires that invokes some emotional reaction could be a valid candi-
identifying these INFLs in what may be a highly segregated date to be investigated.
network. Research in INFL identification is in its very early The applicability of the hypothesis that well-balanced
stages. Some works focus on this task; however, they cur- explanations are more effective than purely factual ones
rently propose approaches for specific domains, such as needs to be verified in different nonpolarized and polarized
marketing [140] and health [183]. Therefore, identifying subjects. The hypothesis is for beliefs surrounding nonpolar-
INFLs on segregated online community networks is an open ized subjects to be less challenging to change. Furthermore,
research area. polarization was shown to have a correlation between multi-
ple subjects. Politics was shown to be a centralizing polariza-
6.2.3. SENTANL Versus Emotion Recognition. SENTANL
tion topic. Therefore, it is expected to be much more
was shown to be the preferred method for belief change
challenging for a polarized individual to change their politi-
detection. Even though SENTANL and emotion recogni-
cal beliefs than their opinion about another polarized subject
tion are sometimes used interchangeably, they are, in fact,
of less centralizing power.
very different. SENTANL identifies the polarity of the per-
son’s attitude toward a given subject as positive, neutral, or
negative. On the other hand, emotion recognition is the 7. Conclusion
task of classifying feelings using an emotion model accord-
This work presented a systematic literature review of online
ing to the psychology of emotions theory. This is a much
belief change from the perspective of three dimensions:
more challenging goal and constitutes an entire subfield
domain-specific user studies, polarization, and opinion
of affective computing [184]. Emotion detection could be
dynamics. We showed evidence that PSOC constructs and
applied automatically to detect the emotions a given expla-
emotional arousal are the two main drivers of online belief
nation elicits in a given user to provide feedback for
change. It was presented that this finding is in line with psy-
improving the explanation generation process. SENTANL
chology research and that due to the close relationship of
can be applied to verify whether a given explanation chan-
individuality with psychological constructs and emotion,
ged the belief and then cross-referenced to the emotion
individuality is arguably the single most influential force in
evoked by the explanation for a deeper insight into the
online belief change. This finding validates the main hypoth-
belief-changing process.
esis of this work, which states that personalization of fake
6.2.4. Polarization Prevention. It was shown in this work that news explanations is a needed improvement for fake news
initially fragmented networks evolve into subnetworks of systems. It was also shown that all the identified cross-
consensus and then into polarized networks. Changing indi- domain trends are rooted in individuality, demonstrating
viduals’ beliefs in a fragmented network may be easier than the importance of personalization to changing preconceived
in polarized networks. For this to happen, the belief- beliefs in fake news. Chiefly, the conclusion was that well-
changing system should attempt to prevent the network balanced explanations between facts and emotionally evok-
from becoming polarized by preempting the network’s natu- ing content that can nudge people into a reflective state are
ral evolution. A preempting belief-changing system would the best candidates for the task. We also presented reasons
be required to include what is known as real-time fake news why these types of explanations may be successful across
detection systems. Real-time methods are receiving the com- multiple fake news domains.
munity’s attention [185]. They aim to identify potentially Polarization was confirmed to be a strong adverse driver
fake news at a speed compatible with the typical fake news of belief change. We have shown alignment between polari-
spread speed. However, this research field is also in its early zation tendencies and individuality. Entrenchment and
stages, offering several opportunities. backfire effect are two constructs that work against belief
change and become especially strong as a reaction to purely
6.2.5. Well-Balanced Explanations. This work verified that factual explanations contradicting preconceived beliefs.
gentle nudging is more efficient than cold factual counterar- Polarization has been shown to be cross-correlated, with
guments to change online beliefs. The field seems primed for politics arguably being the central polarization pole. Polar-
explanations other than fact-heavy text. Figure 1 proposes ized individuals with a specific political ideology also tend
creative explanations as a potential alternative to fact- to be polarized on other subjects such as climate change,
immigration, policy, COVID-19 response, minority rights, [5] Y. Mirsky and W. Lee, “The creation and detection of deep-
and other sensitive topics. Trust is one of the strongest psy- fakes: a survey,” ACM Computing Surveys (CSUR), vol. 54,
chological drivers of polarization. For this reason, INFLs no. 1, pp. 1–41, 2021.
become critical agents of polarization, especially the ones [6] A. Satariano and P. Mozur, “The people onscreen are fake.
who purposely manipulate information for some form of The disinformation is real,” 2023, https://www.nytimes.
personal gain. Furthermore, it was concluded that segregated com/2023/02/07/technology/artificial-intelligence-training-
social networks of opinion evolve through the formation of deepfake.html.
subnetworks of consensus and ultimately to polarized online [7] P. Verma, “They thought loved ones were calling for help. It
social groups. These two findings can potentially be used in was an AI scam,” 2023, https://www.washingtonpost.com/
favor of fake news debunking systems by delivering well- technology/2023/03/05/ai-voice-scam/.
balanced explanations to polarized network INFLs. Since [8] K. Mishima and H. Yamana, “A survey on explainable fake
they are driving forces of opinion formation, changing the news detection,” IEICE Transactions on Information and Sys-
tems, vol. E105.D, no. 7, pp. 1249–1257, 2022.
preconceived beliefs of just a few of these agents may create
a favorable snowball effect in the entire social network [9] A. Raj and M. P. Goswami, “Is fake news spreading more
rapidly than COVID-19 in India,” Journal of Content, Com-
toward consensus against fake news.
munity and Communication, vol. 11, no. 10, pp. 208–220,
2020.
Data Availability Statement [10] R. Denaux, M. Mensio, J. M. Gomez-Perez, and H. Alani,
“Weaving a semantic web of credibility reviews for explain-
The findings supporting this systematic review are from pre- able misinformation detection (extended abstract),” in Pro-
viously reported studies and datasets, which have been cited. ceedings of the Thirtieth International Joint Conference on
The processed data are available from the corresponding Artificial Intelligence, p. 1, Montreal, Canada, August 2021.
author upon request. [11] F. Yang, S. K. Pentyala, S. Mohseni et al., “Xfake: explainable
fake news detector with visualizations,” in The World Wide
Web Conference, pp. 3600–3604, New York, NY, USA, May
Conflicts of Interest 2019.
The authors declare no conflicts of interest. [12] K. Shu, L. Cui, S. Wang, D. Lee, and H. Liu, “defend: explain-
able fake news detection,” in Proceedings of the 25th ACM
SIGKDD International Conference on Knowledge Discovery
Funding & Data Mining, pp. 395–405, New York, NY, 2019.
[13] D. Gunning, M. Stefik, J. Choi, T. Miller, S. Stumpf, and G. Z.
This research was supported by Fundação para a Ciência Yang, “Xai—explainable artificial intelligence,” Science
e Tecnologia (FCT) through (1) INESC-ID Multiannual Robotics, vol. 4, no. 37, 2019.
Funding with reference UIDB/50021/2020 (doi:10.54499/ [14] A. Adadi and M. Berrada, “Peeking inside the black-box: a
UIDB/50021/2020); (2) Grant 2022.09212.PTDC (XAVIER survey on explainable artificial intelligence (xai),” IEEE
Project) and project UIDB/50021/2020 (doi:10.54499/ Access, vol. 6, pp. 52138–52160, 2018.
UIDB/50021/2020), under the auspices of the UNESCO [15] X. Zhang and A. A. Ghorbani, “An overview of online fake
Chair on AI&VR of the University of Lisbon; and (3) news: characterization, detection, and discussion,” Informa-
CHIST-ERA within the CIMPLE project (CHIST-ERA- tion Processing and Management, vol. 57, no. 2, article
19-XAI-003) which corresponds to the FCT reference 102025, 2020.
CHIST-ERA/0001/2019. [16] X. Chen, P. Tsaparas, J. Lijffijt, and T. De Bie, “Opinion
dynamics with backfire effect and biased assimilation,” PLoS
One, vol. 16, no. 9, article e0256922, 2021.
References
[17] P. Atanasova, J. G. Simonsen, C. Lioma, and I. Augenstein,
[1] Y. M. Rocha, G. A. de Moura, G. A. Desidério, C. H. de Oli- “Generating fact checking explanations,” in Proceedings of
veira, F. D. Lourenço, and L. D. de Figueiredo Nicolete, the 58th Annual Meeting of the Association for Computa-
“The impact of fake news on social media and its influence tional Linguistics, pp. 7352–7364, July 2020.
on health during the COVID-19 pandemic: a systematic [18] M. Ruffin, G. Wang, and K. Levchenko, “Explaining why fake
review,” Journal of Public Health, vol. 31, no. 7, pp. 1007– photos are fake: Does it work?,” Proceedings of the ACM on
1016, 2023. Human-Computer Interaction, vol. 7, no. GROUP, pp. 1–
[2] S. A. García, G. G. García, M. S. Prieto, A. J. M. Guerrero, and 22, 2023.
C. R. Jiménez, “The impact of term fake news on the scientific [19] W. Zhang and B. Y. Lim, “Towards relatable explainable AI
community. Scientific performance and mapping in web of with the perceptual process,” in CHI Conference on Human
science,” Social Sciences, vol. 9, no. 5, p. 73, 2020. Factors in Computing Systems, pp. 1–24, New York, NY, April
[3] K. Sipitanos, “Raising awareness against fake news to protect 2022.
democracy: the myth of islamophobia in Trump’s speech,” [20] N. Bryan-Kinns, C. Ford, A. Chamberlain et al., “Explainable
Social Semiotics, vol. 33, no. 4, pp. 714–730, 2023. AI for the arts: XAIxArts,” in Creativity and Cognition, pp. 1–
[4] A. Mitchell, J. Gottfried, G. Stocking, M. Walker, and 7, New York, NY, USA, June 2023.
S. Fedeli, “Many Americans say made-up news is a critical [21] A. Caraban, E. Karapanos, D. Gonçalves, and P. Campos, “23
problem that needs to be fixed,” Pew Research Center, ways to nudge: a review of technology-mediated nudging in
vol. 5, p. 2019, 2019. human-computer interaction,” in Proceedings of the 2019
CHI Conference on Human Factors in Computing Systems, [40] Y. Wang, M. McKee, A. Torbica, and D. Stuckler, “Systematic
pp. 1–15, New York, NY, USA, May 2019. literature review on the spread of health-related misinforma-
[22] F. Altoe and H. S. Pinto, “Towards a personalized online fake tion on social media,” Social Science and Medicine, vol. 240,
news taxonomy,” in Proceedings of the 31st ACM Conference article 112552, 2019.
on User Modeling, Adaptation and Personalization, pp. 96– [41] K. Tsamakis, D. Tsiptsios, B. Stubbs et al., “Summarising data
105, New York, NY, USA, June 2023. and factors associated with COVID19 related conspiracy the-
[23] S. Colton G. A. Wiggins et al., “Computational creativity: the ories in the first year of the pandemic: a systematic review and
final frontier?,” in Ecai, vol. 12, pp. 21–26, Montpelier, Auto- narrative synthesis,” BMC Psychology, vol. 10, no. 1, p. 244,
mated Computer Machinery, New York, NY, 2012. 2022.
[24] A. Jordanous, “Four pppperspectives on computational crea- [42] L. Iandoli, S. Primario, and G. Zollo, “The impact of group
tivity in theory and in practice,” Connection Science, vol. 28, polarization on the quality of online debate in social media:
no. 2, pp. 194–216, 2016. a systematic literature review,” Technological Forecasting
and Social Change, vol. 170, article 120924, 2021.
[25] D. W. MacKinnon, “Creativity: a multi-faceted phenome-
non,” in Creativity: A Discussion at the Nobel Conference, J. [43] J. Khalid, A. Abbas, R. Akbar et al., “Significance of electronic
D. Roslansky, Ed., pp. 17–32, North-Holland, Amsterdan, word of mouth (e-wom) in opinion formation,” International
1970. Journal of Advanced Computer Science and Applications,
vol. 11, no. 2, 2020.
[26] M. Rhodes, “An analysis of creativity,” The Phi Delta Kappan,
vol. 42, no. 7, pp. 305–310, 1961. [44] H. Xie, M. Zhong, Y. Li, and J. C. S. Lui, “Understanding per-
suasion cascades in online product rating systems: modeling,
[27] C. Lamb, D. G. Brown, and C. L. Clarke, “Evaluating compu-
analysis, and inference,” ACM Transactions on Knowledge
tational creativity: an interdisciplinary tutorial,” ACM Com-
Discovery from Data, vol. 15, no. 3, pp. 1–29, 2021.
puting Surveys (CSUR), vol. 51, pp. 1–34, 2018.
[45] A. Ardèvol‐Abreu, H. G. de Zúñiga, and E. Gámez, “The
[28] M. Riedl, “Weird AI Yankovic: generating parody lyrics,”
influence of conspiracy beliefs on conventional and uncon-
2020, https://arxiv.org/abs/2009.12240.
ventional forms of political participation: the mediating role
[29] S. Sharma, S. Agarwal, T. Suresh, P. Nakov, M. S. Akhtar, and of political efficacy,” British Journal of Social Psychology,
T. Chakraborty, “What do you meme? Generating explana- vol. 59, no. 2, pp. 549–569, 2020.
tions for visual semantic role labelling in memes,” Proceed- [46] C. Ruess, C. P. Hoffmann, S. Boulianne, and K. Heger,
ings of the AAAI Conference on Artificial Intelligence, “Online political participation: the evolution of a concept,”
vol. 37, no. 8, pp. 9763–9771, 2023. Information, Communication and Society, vol. 26, no. 8,
[30] N. Köbis and L. D. Mossink, “Artificial intelligence versus pp. 1495–1512, 2023.
Maya Angelou: Experimental evidence that people cannot [47] M. Galesic, H. Olsson, J. Dalege, T. van der Does, and D. L.
differentiate AI-generated from human-written poetry,” Stein, “Integrating social and cognitive aspects of belief
Computers in Human Behavior, vol. 114, article 106553, 2021. dynamics: towards a unifying framework,” Journal of the
[31] J. Toplyn, “Witscript: a system for generating improvised Royal Society Interface, vol. 18, no. 176, article 20200857,
jokes in a conversation,” 2023, https://arxiv.org/abs/2302. 2021.
02008. [48] S. D. Arora, G. P. Singh, A. Chakraborty, and M. Maity,
[32] Y. Chang, X. Wang, J. Wang et al., “A survey on evaluation of “Polarization and social media: a systematic review and
large language models,” ACM Transactions on Intelligent Sys- research agenda,” Technological Forecasting and Social
tems and Technology, vol. 15, no. 3, pp. 1–45, 2024. Change, vol. 183, article 121942, 2022.
[33] Y. Wang, W. Zhong, L. Li et al., “Aligning large language [49] A.-M. Bliuc, A. Bouguettaya, and K. D. Felise, “Online inter-
models with human: a survey,” 2023, https://arxiv.org/abs/ group polarization across political fault lines: an integrative
2307.12966. review,” Frontiers in Psychology, vol. 12, article 641215, 2021.
[34] P. Quintas and H. S. Pinto, Report on the state of the art on [50] G. Yang, Multidisciplinary Studies in Knowledge and Systems
creative xai, vol. 5.1, Tech. rep., CIMPLE project deliverable, Science, IGI Global, Beijing, China, 2013.
2022. [51] N. Van Eck and L. Waltman, “Software survey: Vosviewer, a
[35] H. Noorazar, K. R. Vixie, A. Talebanpour, and Y. Hu, “From computer program for bibliometric mapping,” Scientomet-
classical to modern opinion dynamics,” International Journal rics, vol. 84, no. 2, pp. 523–538, 2010.
of Modern Physics C, vol. 31, no. 7, article 2050101, 2020. [52] R. A. Varanasi, J. Pal, and A. Vashistha, “Accost, accede, or
[36] H. Noorazar, “Recent advances in opinion propagation amplify: attitudes towards COVID-19 misinformation on
dynamics: a 2020 survey,” The European Physical Journal WhatsApp in India,” in CHI Conference on Human Factors
Plus, vol. 135, p. 521, 2020. in Computing Systems, pp. 1–17, New York, NY, April 2022.
[37] R. Ureña, G. Kou, Y. Dong, F. Chiclana, and E. Herrera- [53] T. J. Padamsee, R. M. Bond, G. N. Dixon et al., “Changes in
Viedma, “A review on trust propagation and opinion dynam- COVID-19 vaccine hesitancy among Black and White indi-
ics in social networks and group decision making frame- viduals in the US,” JAMA Network Open, vol. 5, no. 1, article
works,” Information Sciences, vol. 478, pp. 461–475, 2019. e2144470, 2022.
[38] L. Mastroeni, P. Vellucci, and M. Naldi, “Agent-based models [54] T. Do, D. Nguyen, A. Le et al., “Understanding public opinion
for opinion formation: a bibliographic survey,” IEEE Access, on using hydroxychloroquine for COVID-19 treatment via
vol. 7, pp. 58836–58848, 2019. social media,” 2022, http://arxiv.org/abs/2201.00237.
[39] O. Abid, S. Jamoussi, and Y. B. Ayed, “Deterministic models [55] T. Chen, L. Peng, J. Yang, and G. Cong, “Modeling, simulation,
for opinion formation through communication: a survey,” and case analysis of COVID19 over network public opinion
Online Social Networks and Media, vol. 6, pp. 1–17, 2018. formation with individual internal factors and external
information characteristics,” Concurrency and Computation: political orientation and anxiety,” The Social Science Journal,
Practice and Experience, vol. 33, no. 17, article e6201, 2021. vol. 57, pp. 1–10, 2020.
[56] R. Zhang, Y. Shen, C. Lin, and H. Li, “Evolution of public [70] C. Rafkin, A. Shreekumar, and P.-L. Vautrey, “When guid-
opinion on COVID-19 based on microblog visualization,” ance changes: government stances and public beliefs,” Jour-
in 2022 the 5th International Conference on Information Sci- nal of Public Economics, vol. 196, article 104319, 2021.
ence and Systems, pp. 142–148, New York, NY, USA, August [71] M. M. Bowers and C. T. Whitley, “What drives support for
2022. transgender rights? Assessing the effects of biological attribu-
[57] G. Klaus, A. Ernst, and L. Oswald, “Psychological factors tion on U.S. public opinion of transgender rights,” Sex Roles,
influencing laypersons’ acceptance of climate engineering, vol. 83, no. 7-8, pp. 399–411, 2020.
climate change mitigation and business as usual scenarios,” [72] M. O. Ukonu, L. I. Anorue, U. Ololo, and H. M. Olawoyin,
Technology in Society, vol. 60, article 101222, 2020. “Climate of conformism: social media users’ opinion on
[58] T. J. B. Cann, I. S. Weaver, and H. T. P. Williams, “Ideological homosexuality in Nigeria,” SAGE Open, vol. 11, no. 3, article
biases in social sharing of online information about climate 215824402110407, 2021.
change,” PLoS One, vol. 16, no. 4, article e0250656, 2021. [73] A. Freedle and S. Kashubeck-West, “Core belief challenge,
[59] A. Shehata, J. Johansson, B. Johansson, and K. Andersen, rumination, and posttraumatic growth in women following
“Climate change frame acceptance and resistance: extreme pregnancy loss,” Psychological Trauma: Theory, Research,
weather, consonant news, and personal media orientations,” Practice, and Policy, vol. 13, no. 2, pp. 157–164, 2021.
Mass Communication and Society, vol. 25, no. 1, pp. 51–76, [74] S. Lee, L. Atkinson, and Y. H. Sung, “Online bandwagon
2022. effects: quantitative versus qualitative cues in online com-
[60] P. P. Sun, “Understanding EFL university teachers’ synchro- ments sections,” New Media and Society, vol. 24, no. 3,
nous online teaching belief change,” Language Teaching pp. 580–599, 2022.
Research, vol. 1, article 136216882210938, 2022. [75] W. S. Rossi, J. W. Polderman, and P. Frasca, “The closed loop
[61] R. C. Seoane and J. E. Jiménez, “Effectiveness of online and between opinion formation and personalized recommenda-
blended delivery methods on preservice teachers’ knowledge tions,” IEEE Transactions on Control of Network Systems,
and beliefs for writing instruction,” Journal of Education for vol. 9, no. 3, pp. 1092–1103, 2022.
Teaching, vol. 48, no. 2, pp. 178–196, 2022. [76] D. Barman and O. Conlan, “Exploring the links between per-
[62] A. E. Flanigan, M. S. Peteranetz, D. F. Shell, and L.-K. Soh, sonality traits and susceptibility to disinformation,” in Pro-
“Shifting beliefs in computer science: change in CS student ceedings of the 32st ACM Conference on Hypertext and
mindsets,” ACM Transactions on Computing Education, Social Media, pp. 291–294, New York, NY, August 2021.
vol. 22, no. 2, pp. 1–24, 2022. [77] A. H. Y. Chan, R. Horne, H. Lycett et al., “Changing patient
[63] A. Best, “Primary school teachers’ beliefs on computer sci- and public beliefs about antimicrobials and antimicrobial
ence as a discipline and as a school subject,” in Proceedings resistance (AMR) using a brief digital intervention,” Frontiers
of the 15th Workshop in Primary and Secondary Computing in Pharmacology, vol. 12, article 608971, 2021.
Education, WiPSCE 2020, vol. 1, p. 1, Association for Com- [78] I.’. I. L. D. Pinto, N. Rungratsameetaweemana, K. Flaherty
puting Machinery, 2020. et al., “Intermittent brain network reconfigurations and the
[64] L. L. Gelauff and A. Goel, “Opinion change or differential resistance to social media influence,” Network Neuroscience,
turnout: Austin’s budget feedback exercise and the police vol. 6, no. 3, pp. 870–896, 2022.
department,” in Equity and Access in Algorithms, Mecha- [79] M. Vlasceanu, M. J. Morais, and A. Coman, “Network struc-
nisms, and Optimization, p. 1, New York, NY, USA, October ture impacts the synchronization of collective beliefs,” Journal
2022. of Cognition and Culture, vol. 21, no. 5, pp. 431–448, 2021.
[65] Y. Tsfati, J. Cohen, S. Dvir-Gvirsman, K. Tsuriel, I. Waismel- [80] V. Grimm and F. Mengel, “Experiments on belief formation
Manor, and R. L. Holbert, “Political para-social relationship in networks,” Journal of the European Economic Association,
as a predictor of voting preferences in the Israeli 2019 elec- vol. 18, no. 1, pp. 49–82, 2020.
tions,” Communication Research, vol. 49, no. 8, pp. 1118– [81] M. Xu, Z. Luo, R. Liu, B. Wang, and H. Xu, “One-sided versus
1147, 2022. two-sided: a novel opinion dynamics information-type
[66] J.-P. Fränken, N. Theodoropoulos, A. Moore, and N. Bramley, education-based Hegselmann–Krause model,” in 2021 IEEE
“Belief revision in a micro-social network: modeling sensitivity International Conference on Systems, Man, and Cybernetics
to statistical dependencies in social learning,” CogSci, vol. 1, (SMC), p. 339-334, Melbourne, Australia, 2021.
pp. 1255–1261, 2020. [82] A. Mansouri and F. Taghiyareh, “Effect of segregation on
[67] J. Massachs, C. Monti, G. D. F. De Francisci Morales, and opinion formation in scale-free social networks: an agent-
F. Bonchi, “Roots of Trumpism: homophily and social feed- based approach,” International Journal of Engineering,
back in Donald Trump support on Reddit,” in 12th ACM vol. 34, no. 1, pp. 66–74, 2021.
Conference on Web Science, Association for Computing [83] Y. Peng, Y. Zhao, and J. Hu, “On the role of community
Machinery, pp. 49–58, New York, NY, USA, July 2020. structure in evolution of opinion formation: a new bounded
[68] R. J. Garcia, E. V. Shaw, and N. Scurich, “Normative and confidence opinion dynamics,” Information Sciences, vol. 621,
informational influence in group decision making: effects of pp. 672–690, 2023.
majority opinion and anonymity on voting behavior and [84] H. Takesue, “A noisy opinion formation model with two
belief change,” Group Dynamics: Theory, Research, and Prac- opposing mass media,” 2020, https://arxiv.org/abs/2011.
tice, vol. 25, no. 4, pp. 319–333, 2021. 13813.
[69] F. Rigoli, “Opinions about immigration, patriotism, and wel- [85] L. Burbach, P. Halbach, M. Ziefle, and A. Calero Valdez,
fare policies during the coronavirus emergency: the role of “Opinion formation on the internet: the influence of
personality, network structure, and content on sharing mes- [100] Z. Li, X. Tang, and Z. Hong, “Opinion dynamics induced by
sages online,” Frontiers in Artificial Intelligence, vol. 3, p. 45, agents with particular goal,” Journal of Systems Science and
2020. Complexity, vol. 35, no. 6, pp. 2319–2335, 2022.
[86] T. Chen, X. Yin, J. Yang, G. Cong, and G. Li, “Modeling [101] I. V. Kozitsin, “Formal models of opinion formation and their
multi-dimensional public opinion process based on complex application to real data: evidence from online social net-
network dynamics model in the context of derived topics,” works,” The Journal of Mathematical Sociology, vol. 46,
Axioms, vol. 10, no. 4, p. 270, 2021. no. 2, pp. 120–147, 2022.
[87] M. Bashari and M.-R. Akbarzadeh-T, “Theoretical develop- [102] I. V. Pasquetto, E. Jahani, S. Atreja, and M. Baum, “Social
ment of a probabilistic fuzzy model for opinion formation debunking of misinformation on WhatsApp: the case for
in social networks,” Fuzzy Sets and Systems, vol. 454, strong and in-group ties,” Proceedings of the ACM on
pp. 125–148, 2023. Human-Computer Interaction, vol. 6, no. CSCW1, pp. 1–35,
[88] Z. Wu, Q. Zhou, Y. Dong, J. Xu, A. H. Altalhi, and F. Herrera, 2022.
“Mixed opinion dynamics based on DeGroot model and [103] Y. Tang, J. Liu, and W. Chen, “Exchange, adopt, evolve:
Hegselmann–Krause model in social networks,” IEEE Trans- modeling the spreading of opinions through cognition and
actions on Systems, Man, and Cybernetics: Systems, vol. 53, interaction in a social network,” Information Sciences,
no. 1, pp. 296–308, 2023. vol. 551, pp. 1–22, 2021.
[89] Z. Zhang, S. Al-Abri, and F. Zhang, “Opinion dynamics on [104] H.-J. Geiss, F. Sakketou, and L. Flek, “Ok boomer: probing the
the sphere for stable consensus and stable bipartite dissen- socio-demographic divide in echo chambers,” in Proceedings
sus,” IFAC-PapersOnLine, vol. 55, no. 13, pp. 288–293, 2022. of the Tenth International Workshop on Natural Language
[90] M. Bashari and M.-R. Akbarzadeh-T, “Controlling opinions Processing for Social Media, pp. 83–105, Seattle, Washington,
in Deffuant model by reconfiguring the network topology,” July 2022.
Physica A: Statistical Mechanics and its Applications, vol. 544, [105] T. Ha, “Understanding of majority opinion formation in
article 123462, 2020. online environments through statistical analysis of news,
[91] I. V. Kozitsin, “A general framework to link theory and documentary, and comedy YouTube channels,” Social Sci-
empirics in opinion formation models,” Scientific Reports, ence Computer Review, vol. 41, no. 2, pp. 353–369, 2022.
vol. 12, no. 1, p. 5543, 2022. [106] C. Monti, G. De Francisci Morales, and F. Bonchi, “Learning
[92] J. Wei, Y. Jia, Y. Zhang, H. Zhu, and W. Huang, “The public opinion dynamics from social traces,” in Proceedings of the
opinion evolution under group interaction in different infor- 26th ACM SIGKDD International Conference on Knowledge
mation features,” Complexity, vol. 2022, Article ID 1016692, Discovery & Data Mining, pp. 764–773, New York,
15 pages, 2022. NY, USA, August 2020.
[93] A. Mansouri and F. Taghiyareh, “Phase transition in the [107] J. Ding, M. Xu, Y. K. Tse, K. Y. Lin, and M. Zhang, “Customer
social impact model of opinion formation in log-normal net- opinions mining through social media: insights from sustain-
works,” Journal of Information Systems and Telecommunica- ability fraud crisis-Volkswagen emissions scandal,” Enterprise
tion (JIST), vol. 9, no. 33, pp. 1–14, 2021. Information Systems, vol. 17, no. 8, article 2130012, 2023.
[94] M. Liu and L. Rong, “An online multi-dimensional opinion [108] T. Rudnyk and O. Chertov, “Method for identifying twitter
dynamic model with misinformation diffusion in emergency accounts that have changed their opinion about politicians,”
events,” Journal of Information Science, vol. 48, no. 5, in ITS, pp. 24–35, CEUR-WS, Kyiev, Ukraine, 2020.
pp. 640–659, 2022. [109] E. Biondi, C. Boldrini, A. Passarella, and M. Conti, “Dynam-
[95] Q. Zhou and Z. Wu, “Multidimensional Friedkin-Johnsen ics of opinion polarization,” IEEE Transactions on Systems,
model with increasing stubbornness in social networks,” Man, and Cybernetics: Systems, vol. 53, no. 9, pp. 5381–
Information Sciences, vol. 600, pp. 170–188, 2022. 5392, 2023.
[96] Y. Luo, C. Cheng, Y. Li, and C. Yu, “Opinion formation with [110] W. Xu, L. Zhu, J. Guan, Z. Zhang, and Z. Zhang, “Effects of
zealots on temporal network,” Communications in Nonlinear stubbornness on opinion dynamics,” in Proceedings of the
Science and Numerical Simulation, vol. 98, article 105772, 2021. 31st ACM International Conference on Information &
[97] Y. Li, Z. Chen, and H. V. Zhao, “Robust opinion control Knowledge Management, pp. 2321–2330, New York, NY,
under network perturbation,” IEEE Signal Processing Letters, October 2022.
vol. 29, pp. 1649–1653, 2022. [111] T. Kinoshita and M. Aida, “A spectral-based model for
[98] S. Cheng, C. Sun, S. Yang, M. Xu, and H. Xu, “Jumping on the describing social polarization in online communities,” IEICE
bandwagon: group opinion prompts agents to reach consen- Transactions on Communications, vol. E105.B, no. 10,
sus,” in 2021 IEEE 23rd Int Conf on High Performance Com- pp. 1181–1191, 2022.
puting & Communications; 7th Int Conf on Data Science & [112] J. Gaitonde, J. Kleinberg, and É. Tardos, “Polarization in geo-
Systems; 19th Int Conf on Smart City; 7th Int Conf on Depend- metric opinion dynamics,” in Proceedings of the 22nd ACM
ability in Sensor, Cloud & Big Data Systems & Application Conference on Economics and Computation, pp. 499–519,
(HPCC/DSS/SmartCity/DependSys), pp. 399–404, Haikou, New York, NY, July 2021.
Hainan, China, December 2021. [113] H. P. Maia, S. C. Ferreira, and M. L. Martins, “Adaptive net-
[99] S. Gündüç, “The effect of social media on shaping individuals work approach for emergence of societal bubbles,” Physica A:
opinion formation,” in Complex Networks and their Applica- Statistical Mechanics and its Applications, vol. 572, article
tions VIII: Volume 2 Proceedings of the Eighth International 125588, 2021.
Conference on Complex Networks and their Applications [114] H. A. Prasetya and T. Murata, “A model of opinion and prop-
COMPLEX NETWORKS 2019 8, pp. 376–386, Springer, Pen- agation structure polarization in social media,” Computa-
silvania, PA, 2019. tional Social Networks, vol. 7, no. 1, pp. 1–35, 2020.
[115] T. Donkers and J. Ziegler, “The dual echo chamber: modeling [130] P. V. Sheela and F. Mannering, “The effect of information on
social media polarization for interventional recommending,” changing opinions toward autonomous vehicle adoption: an
in Fifteenth ACM Conference on Recommender Systems, exploratory analysis,” International Journal of Sustainable
pp. 12–22, New York, NY, September 2021. Transportation, vol. 14, no. 6, pp. 475–487, 2020.
[116] L. Wang, C. A. Wang, and X. Yao, “Befriended to polarise? [131] S. M. Coles and M. Saleem, “Social media expression and user
The impact of friend identity on review polarisation—A- predispositions: applying the differential susceptibility to
quasi‐experiment,” Information Systems Journal, vol. 1, media effects model to the study of issue polarization,” Social
2023. Media + Society, vol. 7, no. 4, article 205630512110529, 2021.
[117] H. Min, J. Cao, J. Ge, and B. Liu, “A multi-agent system for [132] A. Tyagi, J. Uyheng, and K. M. Carley, “Affective polarization
fine-grained opinion dynamics analysis in online social net- in online climate change discourse on twitter,” in 2020 IEEE/
works,” IEEE Transactions on Computational Social Systems, ACM International Conference on Advances in Social Net-
vol. 11, no. 1, pp. 815–828, 2022. works Analysis and Mining (ASONAM), pp. 443–447, The
[118] C. S. R. Avuthu, M. Maleszka, and N. Van Sinh, “Inter- Hague, Netherlands, December 2020.
changeability of knowledge and opinion integration strategies [133] A. Spitz, A. Abu-Akel, and R. West, “Interventions for soften-
in collective models,” in 2020 IEEE International Conference ing can lead to hardening of opinions: evidence from a ran-
on Systems, Man, and Cybernetics (SMC), pp. 2196–2200, domized controlled trial,” in Proceedings of the Web
Toronto, ON, Canada, October 2020. Conference 2021no. p, pp. 1098–1109, New York, NY, USA,
[119] N. Hirakura, M. Aida, and K. Kawashima, “Modeling polari- April 2021.
zation caused by empathetic and repulsive reaction in online [134] N. Yeung, J. Lai, and J. Luo, “Face off: polarized public opin-
social network,” IEICE Transactions on Communications, ions on personal face mask usage during the COVID-19 pan-
vol. E105.B, no. 8, pp. 990–1001, 2022. demic,” in 2020 IEEE International Conference on Big Data
[120] S. Tu and S. Neumann, “A viral marketing-based model for (Big Data), pp. 4802–4810, Atlanta, GA, USA, December
opinion dynamics in online social networks,” in Proceedings 2020.
of the ACM Web Conference 2022, pp. 1570–1578, New York, [135] S. Perrett, “A divided kingdom? Variation in polarization,
NY, April 2022. sorting, and dimensional alignment among the British public,
[121] H. Ferraz de Arruda, F. Maciel Cardoso, G. Ferraz de Arruda, 1986–2018,” British Journal of Sociology, vol. 72, no. 4,
A. R. Hernández, L. da Fontoura Costa, and Y. Moreno, pp. 992–1014, 2021.
“Modelling how social network algorithms can influence [136] C. Largeron, A. Mardale, and M. A. Rizoiu, “Linking the
opinion polarization,” Information Sciences, vol. 588, dynamics of user stance to the structure of online discus-
pp. 265–278, 2022. sions,” in Computer Science (including subseries Lecture Notes
[122] U. Chitra and C. Musco, “Analyzing the impact of filter in Artificial Intelligence and Lecture Notes in Bioinformatics),
bubbles on social network polarization,” in Proceedings of Lecture Notes in Computer Science ((LNISA,volume 12695)),
the 13th International Conference on Web Search and Data Springer, Pensilvania, PA, 2021.
Mining, pp. 115–123, New York, NY, USA, January 2020. [137] J. J. B. Mijs, W. de Koster, and J. van der Waal, “Belief change
[123] N. Loy, M. Raviola, and A. Tosin, “Opinion polarization in in times of crisis: providing facts about COVID-19-induced
social networks,” Philosophical Transactions of the Royal inequalities closes the partisan divide but fuels intra-
Society A, vol. 380, no. 2224, article 20210158, 2022. partisan polarization about inequality,” Social Science
[124] D. A. Gubanov, I. V. Petrov, and A. G. Chkhartishvili, “Mul- Research, vol. 104, article 102692, 2022.
tidimensional model of opinion dynamics in social networks: [138] G. Fischer, P. Riedesser, and A. G. Fischer, Lehrbuch der
polarization indices,” Automation and Remote Control, psychotraumatologie, Ernst Reinhardt Verlag München,
vol. 82, no. 10, pp. 1802–1811, 2021. München, DE, 2020.
[125] M. W. Macy, M. Ma, D. R. Tabin, J. Gao, and B. K. Szy- [139] A. Boals, “Trauma in the eye of the beholder: objective and
manski, “Polarization and tipping points,” Proceedings of subjective definitions of trauma,” Journal of Psychotherapy
the National Academy of Sciences of the United States of Integration, vol. 28, no. 1, pp. 77–89, 2018.
America, vol. 118, no. 50, 2021. [140] P. Harrigan, T. M. Daly, K. Coussement, J. A. Lee, G. N. Sou-
[126] S. Gupta, G. Jain, and A. A. Tiwari, “Polarised social media tar, and U. Evers, “Identifying influencers on social media,”
discourse during COVID-19 pandemic: evidence from You- International Journal of Information Management, vol. 56,
Tube,” Behaviour & Information Technology, vol. 42, no. 2, article 102246, 2021.
pp. 227–248, 2023. [141] Z. Liberman, K. D. Kinzler, and A. L. Woodward, “Origins of
[127] J. Bernacer, J. García-Manglano, E. Camina, and F. Güell, homophily: Infants expect people with shared preferences to
“Polarization of beliefs as a consequence of the COVID-19 affiliate,” Cognition, vol. 212, article 104695, 2021.
pandemic: The case of spain,” PLoS One, vol. 16, no. 7, article [142] M. Cinelli, G. De Francisci Morales, A. Galeazzi,
e0254511, 2021. W. Quattrociocchi, and M. Starnini, “The echo chamber
[128] S. Nair, A. Iamnitchi, and J. Skvoretz, “Promoting social con- effect on social media,” Proceedings of the National Academy
ventions across polarized networks: an empirical study,” in of Sciences, vol. 118, no. 9, article e2023301118, 2021.
Proceedings of the 2019 IEEE/ACM International Conference [143] R. S. Nickerson, “Confirmation bias: a ubiquitous phenome-
on Advances in Social Networks Analysis and Mining, non in many guises,” Review of General Psychology, vol. 2,
pp. 349–352, New York, NY, USA, August 2019. no. 2, pp. 175–220, 1998.
[129] G. De Francisci Morales, C. Monti, and M. Starnini, “No echo [144] T. T. Nguyen, P.-M. Hui, F. M. Harper, L. Terveen, and J. A.
in the chambers of political interactions on Reddit,” Scientific Konstan, “Exploring the filter bubble: The effect of using rec-
Reports, vol. 11, no. 1, p. 2818, 2021. ommender systems on content diversity,” in Proceedings of
the 23rd international conference on World wide web, [163] P. C. Smith, Promoting Belief Change by Encouraging Evalu-
pp. 677–686, New York, NY, USA, April 2014. ation of Prior Beliefs, The University of Wisconsin-Milwau-
[145] D. M. Rousseau, S. B. Sitkin, R. S. Burt, and C. Camerer, “Not kee, Milwaukee, WI, 2000.
so different after all: a cross-discipline view of trust,” Acad- [164] C. R. Berger and R. J. Calabrese, “Some explorations in initial
emy of Management Review, vol. 23, no. 3, pp. 393–404, 1998. interaction and beyond: toward a developmental theory of
[146] R. F. Baumeister and M. R. Leary, “The need to belong: desire interpersonal communication,” Human Communication
for interpersonal attachments as a fundamental human motiva- Research, vol. 1, no. 2, pp. 99–112, 1975.
tion,” Psychological Bulletin, vol. 117, no. 3, pp. 497–529, 1995. [165] J.-J. Igartua and I. Barrios, “Changing real-world beliefs with
[147] G. D. Bishop and D. G. Myers, “Informational influence in controversial movies: processes and mechanisms of narrative
group discussion,” Organizational Behavior and Human Per- persuasion,” Journal of Communication, vol. 62, no. 3,
formance, vol. 12, no. 1, pp. 92–104, 1974. pp. 514–531, 2012.
[148] E. Noelle-Neumann, The Spiral of Silence: Public Opinion– [166] D. Bao, D. Dong, and X. Meng, “Parasocial interaction
Our Social Skin, University of Chicago Press, Chicago, IL, between browser and poster in virtual communities: an
1993. empirical study on dianping. com,” Chinese Journal of Man-
[149] T.-Y. Wu and C. A. Lin, “Predicting the effects of eWOM and agement, vol. 8, no. 7, pp. 1010–1020, 2011.
online brand messaging: source trust, bandwagon effect and [167] R. S. Solomon, P. Y. K. L. Srinivas, A. Das, B. Gamback, and
innovation adoption factors,” Telematics and Informatics, T. Chakraborty, “Understanding the psycho-sociological
vol. 34, no. 2, pp. 470–480, 2017. facets of homophily in social network communities,” IEEE
[150] K. Tapson, M. Doyle, V. Karagiannopoulos, and P. Lee, Computational Intelligence Magazine, vol. 14, no. 2, pp. 28–
“Understanding moral injury and belief change in the experi- 40, 2019.
ences of police online child sex crime investigators: an inter- [168] H. Huang, “A cross-cultural test of the spiral of silence,”
pretative phenomenological analysis,” Journal of Police and International Journal of Public Opinion Research, vol. 17,
Criminal Psychology, vol. 37, no. 3, pp. 637–649, 2022. no. 3, pp. 324–345, 2005.
[151] L. G. Calhoun and R. G. Tedeschi, Posttraumatic growth: The [169] D. Keltner, K. Oatley, and J. M. Jenkins, Understanding Emo-
positive lessons of loss, American Psychological Association, tions, Wiley Hoboken, NJ, New Jersey, 2014.
2001. [170] E. Diener and R. A. Emmons, “The independence of positive
[152] W. Xu, H. Jiang, Y. Zhou, L. Zhou, and H. Fu, “Intrusive and negative affect,” Journal of Personality and Social Psy-
rumination, deliberate rumination, and posttraumatic chology, vol. 47, no. 5, pp. 1105–1117, 1984.
growth among adolescents after a tornado: the role of social [171] R. J. Larsen and E. Diener, Promises and problems with the
support,” The Journal of Nervous and Mental Disease, circumplex model of emotion, Sage Publications, Inc, 1992.
vol. 207, no. 3, pp. 152–156, 2019. [172] R. R. McCrae and P. T. Costa Jr., “Adding liebe und arbeit:
[153] A. M. Brandt, “Racism and research: the case of the Tuskegee the full five-factor model and well-being,” Personality and
syphilis study,” Hastings Center Report, vol. 8, no. 6, pp. 21– Social Psychology Bulletin, vol. 17, no. 2, pp. 227–232, 1991.
29, 1978. [173] C. D. Broad, “Emotion and sentiment,” The Journal of Aes-
[154] S. Zielinski, “Henrietta lacks ‘immortal’cells,” Smithsonian thetics and Art Criticism, vol. 13, no. 2, pp. 203–214, 1954.
Magazine, vol. 22, 2010. [174] A. Tellegen, Personality traits: Issues of definition, evidence,
[155] H. M. Boehme, R. J. Kaminski, and M. S. Nolan, “City-wide and assessment, University of Minnesota Press, 1991.
firearm violence spikes in Minneapolis following the murder [175] J. J. Gross, S. K. Sutton, and T. Ketelaar, “Relations between
of George Floyd: a comparative time-series analysis of three affect and personality: support for the affect-level and
cities,” Urban Science, vol. 6, no. 1, p. 16, 2022. affective-reactivity views,” Personality and Social Psychology
[156] N. Schwarz and G. L. Clore, “Mood as information: 20 years Bulletin, vol. 24, no. 3, pp. 279–288, 1998.
later,” Psychological Inquiry, vol. 14, no. 3, pp. 296–303, 2003. [176] D. Watson and L. A. Clark, “Negative affectivity: the disposi-
[157] R. B. Cialdini and N. J. Goldstein, “Social influence: compli- tion to experience aversive emotional states,” Psychological
ance and conformity,” Annual Review of Psychology, vol. 55, Bulletin, vol. 96, no. 3, pp. 465–490, 1984.
no. 1, pp. 591–621, 2004. [177] F. Rogers, “Personality and individuality,” The North Ameri-
[158] K. H. Jamieson and J. N. Cappella, Echo Chamber: Rush Lim- can Review, vol. 214, no. 791, pp. 514–517, 1921.
baugh and the Conservative Media Establishment, Oxford [178] L. Yue, W. Chen, X. Li, W. Zuo, and M. Yin, “A survey of sen-
University Press, Oxford, EN, 2008. timent analysis in social media,” Knowledge and Information
[159] E. Pariser, The filter bubble: what the Internet is hiding from Systems, vol. 60, no. 2, pp. 617–663, 2019.
you, Penguin UK, London, England, 2011. [179] Y. Wang, Z. Chen, and C. Fu, “Synergy masks of domain
[160] A. J. Stewart, M. Mosleh, M. Diakonova, A. A. Arechar, D. G. attribute model DaBERT: emotional tracking on time-
Rand, and J. B. Plotkin, “Information gerrymandering and varying virtual space communication,” Sensors, vol. 22,
undemocratic decisions,” Nature, vol. 573, no. 7772, no. 21, p. 8450, 2022.
pp. 117–121, 2019. [180] K. Vishnubhotla and S. M. Mohammad, “Tweet emotion
[161] A. Melinder, T. Brennen, M. F. Husby, and O. Vassend, dynamics: emotion word usage in tweets from US and Can-
“Personality, confirmation bias, and forensic interviewing ada,” 2022, https://arxiv.org/abs/2204.04862.
performance,” Applied Cognitive Psychology, vol. 34, no. 5, [181] S. Mohammad, “Obtaining reliable human ratings of valence,
pp. 961–971, 2020. arousal, and dominance for 20,000 English words,” in Pro-
[162] B. De Raad, The Big Five Personality Factors: The Psycholexi- ceedings of the 56th annual meeting of the association for com-
cal Approach to Personality, Hogrefe & Huber Publishers, putational linguistics (volume 1: Long papers), pp. 174–184,
Newburyport, MA, 2000. Melbourne, Australia, July 2018.
[182] R. Krishna and C. M. Prashanth, “Finding context-based

influencers on Twitter,” SN Computer Science, vol. 5, no. 1,
2024.
[183] R. O. Lutkenhaus, J. Jansz, and M. P. A. Bouman, “Tailoring
in the digital era: stimulating dialogues on health topics in
collaboration with social media influencers,” Digital Health,
vol. 5, article 205520761882152, 2019.
[184] S. B. Daily, M. T. James, D. Cherry et al., “Affective comput-
ing: historical foundations, current applications, and future
trends,” Emotions and Affect in Human Factors and
Human-Computer Interaction, vol. 1, pp. 213–231, 2017.
[185] C. Zhang, A. Gupta, X. Qin, and Y. Zhou, “A computational
approach for real-time detection of fake news,” Expert Sys-
tems with Applications, vol. 221, article 119656, 2023.
[186] D. Matravers, Art and Emotion, Oxford University Press,
Oxford, England, 2001.
[187] Y. Treadwell, “Humor and creativity,” Psychological Reports,
vol. 26, no. 1, pp. 55–58, 1970.
[188] J. Morreall, “Humor and emotion,” American Philosophical
Quarterly, vol. 20, no. 3, pp. 297–304, 1983.

Altoe Et Al., 2024

Uploaded by

Copyright:

Available Formats

Altoe Et Al., 2024

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Altoe Et Al., 2024

Uploaded by

Copyright:

Available Formats

Hindawi

Human Behavior and Emerging Technologies

Filipe Altoe ,1 Catarina Moreira ,2 H. Sofia Pinto ,1 and Joaquim A. Jorge 1

Correspondence should be addressed to Filipe Altoe; [email protected]

Academic Editor: Mirko Duradoni

1. Introduction artifacts, increasing the diﬃculty for news consumers to

Figure 1: Belief change comparison: fact-based and CC-generated creative explanations.

Table 1: Inclusion and exclusion criteria.

Domains User intervention

Figure 2: Opinion and belief change 2020–2023.

Figure 3: Keyword combinations.

Table 2: Taxonomy of included papers in review.

6. Discussion Table 3 shows that four cross-dimensional trends were

Table 3: Reviewed papers addressing each cross-dimensional trend.

SUBCNS FACBAL INFL SENTANL PSOC CRSPOL ERLPUB EMLNDG Total

[182] R. Krishna and C. M. Prashanth, “Finding context-based

You might also like