Fake News Detection

The emergence of various social networks has generated vast volumes of data. Efficient
methods for capturing, distinguishing, and filtering real and fake news are becoming
increasingly important, especially after the outbreak of the COVID-19 pandemic. This study
conducts a multi-aspect and systematic review of the current state and challenges of graph
neural networks (GNNs) for fake news detection systems and outlines a comprehensive
approach to implementing fake news detection systems using GNNs. Furthermore, advanced
GNN-based techniques for implementing pragmatic fake news detection systems are discussed
from multiple perspectives. First, we introduce the background and overview related to fake
news, fake news detection, and GNNs. Second, we provide a GNN taxonomy-based fake news
detection taxonomy and review and highlight models in categories. Subsequently, we compare
critical ideas, advantages, and disadvantages of the methods in categories. Next, we discuss the
possible challenges of fake news detection and GNNs. Finally, we present several open issues in
this area and discuss potential directions for future research. We believe that this review can be
utilized by systems practitioners and newcomers in surmounting current impediments and
navigating future situations by deploying a fake news detection system using GNNs.
Keywords: Fake news, Fake news characteristics, Fake news features, Fake news detection,
Graph neural network

1. Introduction

Recently, social networks have contributed to an explosion of information. Social networks

have become the main communication channel for people worldwide. However, the veracity of
news posted on social networks often cannot be determined. Thus, using social networks is a
double-edged sword. Therefore, if the news received from social networks is real, it will be
beneficial. Conversely, if this news is fake, it will have many harmful consequences, and the
extent of damage when fake news is widely disseminated is incalculable.

Fake news is a source of absolutely inventive information to spread deceptive content or

entirely misrepresent actual news articles [1]. Numerous examples of fake news exist. Hunt
et al. [2] indicated that during the 2016 US presidential election, the activity of Clinton
supporters was affected by the spread of traditional center and left-leaning news from top
influencers, whereas the movement of Trump supporters was influenced by the dynamics of
top fake news spreaders. Moreover, public opinion manipulation based on the spread of fake
news related to the Brexit vote in the United Kingdom was reported. Most recently, the
prevalence of fake news has been witnessed during the COVID-19 pandemic. These examples
show that the spread of fake news on social networks has a significant effect on many fields.
Timely detection and containment of fake news before widespread dissemination is an urgent
task. Therefore, many methods have been implemented to detect and prevent the spread of
fake news over the past decade, among which the graph neural network (GNN)-based approach
is the most recent.
Based on previous studies’ findings regarding the benefit of using GNNs for fake news
detection, we summarize some main justifications for using GNNs as follows. Existing
approaches for fake news detection focus almost exclusively on features related to the content,
propagation, and social context separately in their models. GNNs promise to be a potentially
unifying framework for combining content, propagation, and social context-based
approaches [3]. Fake news spreaders can attack machine learning-based models because these
models depend strongly on news text. Making detection models less dependent on the news
text is necessary to avoid this issue. GNN-based models can achieve similar or higher
performance than modern methods without textual information [4]. GNN-based approaches
can provide flexibility in defining the information propagation pattern using parameterized
random walks and iterative aggregators [5].

A graph neural network is a novel technique that focuses on using deep learning algorithms
over graph structures [6]. Before their application in fake news detection systems, GNNs had
been successfully applied in many machine learning and natural language processing-related
tasks, such as object detection [7], [8], sentiment analysis [9], [10], and machine
translation [11], [12]. The rapid development of numerous GNNs has been achieved by
improving convolutional neural networks, recurrent neural networks, and autoencoders
through deep learning [13]. The rapid development of GNN-based methods for fake news
detection systems on social networks can be attributed to the rapid growth of social networks
in terms of the number of users, the amount of news posted, and user interactions.
Consequently, social networks naturally become complex graph structures if they are applied
independently, which is problematic for previous machine learning-based and deep learning-
based fake news detection algorithms. The main reasons for this phenomenon are the
dependence of the graph size on the number of nodes and the different numbers of node
neighbors. Therefore, some important operations (convolutions) are difficult to calculate in the
graph domain. Additionally, the primary assumption of previous machine learning and deep
learning-based fake news detection algorithms is that news items are independent. This
assumption cannot apply to graph data because nodes can connect to other nodes through
various types of relationships, such as citations, interactions, and friendships. GNN-based fake
news detection methods have been developed. Although some state-of-the-art results have
been achieved (see Table 1), no complete GNN-based fake news detection and prevention
system existed when we conducted this study. Faking news on social networks is still a major
challenge that needs to be solved (the first justification).
Table 1
A description of the improved performance of the traditional methods compared using GNN-
based methods.

Least improved
Method Ref Improved methods Dataset
performance
DTC, SVM-TS, mGRU, RFC, Twitter15, Accuracy: 18.7%,
GCAN [14]
tCNN, CRNN, CSI, dEFEND Twitter16 Accuracy: 19.9%
FANG [5] Feature SVM, CSI Twitter AUC: 6.07%
HAN, dEFEND, SAFE, CNN, FakeNewsNet, F1: 5.19%,
SAFER [15]
RoBERTa, Maj sharing baseline FakeHealth F1: 5.00%
Weibo, Accuracy: 4.5%,
DTC, SVM-RBF, SVM-TS, RvNN
Bi-GCN [16] Twitter15, Accuracy: 13.6%,
PPC_RNN+CNN
Twitter16 Accuracy: 14.3%
SVM, LIWC, text-CNN, Label
AA- PolitiFact Accuracy: 2.82%
[17] propagation,
HGNN BuzzFeed Accuracy:9.34%
DeepWalk, LINE, GAT, GCN, HAN
Open in a separate window

Various survey papers of fake news detection have been published, such
as [18], [19], [20], [21], [22], [23]. We briefly summarize related work as follows: Vitaly Klyuev
et al. [20] presented a survey of different fake news detection methods based on semantics
using natural language processing (NLP) and text mining techniques. Additionally, the authors
discussed automatic checking and bot detection on social networks. Meanwhile, Oshikawa
et al. [21] introduced a survey for fake news detection, focusing only on reviewing NLP-based
approaches. Collins et al. [18] presented various variants of fake news and reviewed recent
trends in preventing the spread of fake news on social networks. Shu et al. [22] conducted a
review on various types of disinformation, factor influences, and approaches that decrease the
effects. Khan et al. [19] presented fake news variants, such as misinformation, rumors, clickbait,
and disinformation. They provided a more detailed representation of some fake news variant
detection methods without limiting NLP-based approaches. They also introduced types of
available detection models, such as knowledge-based, fact-checking, and hybrid approaches.
Moreover, the authors introduced governmental strategies to prevent fake news and its
variants. Mahmud et al. [23] presented a comparative analysis by implementing several
commonly used methods of machine learning and GNNs for fake news detection on social
media and comparing their performance. No survey papers have attempted to provide a
comprehensive and thorough overview of fake news detection using the most current
technique, namely, the GNN-based approach (the second justification).
The above two justifications motivated us to conduct this survey. Although some similarities are
unavoidable, our survey is different from the aforementioned works in that we focus on
description, analysis, and discussion of the models of fake news detection using the most
recent GNN-based techniques. We believe that this paper can provide an essential and basic
reference for new researchers, newcomers, and systems practitioners in overcoming current
barriers and forming future directions when improving the performance of fake news detection
systems using GNNs. This paper makes the following four main contributions.

We provide the most comprehensive survey yet of fake news, including similar concepts,
characteristics, types of related features, types of approaches, and benchmarks datasets. We
redefine similar concepts regarding fake news based on their characteristics. This survey can
serve as a practical guide for elucidating, improving, and proposing different fake news
detection methods.

We provide a brief review of existing types of GNN models. We also make necessary
comparisons among types of models and summarize the corresponding algorithms.

We introduce the details of GNN models for fake news detection systems, such as pipelines of
models, benchmark datasets, and open source code. These details provide a background and
guide experienced developers in proposing different GNNs for fake news prevention
applications.

We introduce and discuss open problems for fake news detection and prevention using GNN
models. We provide a thorough analysis of each issue and propose future research directions
regarding model depth and scalability trade-offs.

This section justified the problem and highlighted our motivations for conducting this survey.
The remaining sections of the paper are ordered as follows. Section 2 introduces the
background and provides an overview of fake news, fake news detection, and GNNs.
Section 3 presents the survey methodology used to conduct the review. General information on
the included papers is analyzed in Section 4. In Section 5, the selected papers are categorized
and reviewed in detail. Subsequently, we discuss the comparisons, advantages, and
disadvantages of the methods by category in Section 6. Next, the possible challenges of fake
news and GNNs are briefly evaluated in Section 7. Finally, we identify several open issues in this
area and discuss potential directions for future research in Section 8.

2. Background

2.1. Understanding fake news

What is fake news? News is understood as meta-information and can include the
following [24]:
Source: Publishers of news, such as authors, websites, and social networks.

Headline: Description of the main topic of the news with a short text to attract readers’
attention.

Body content: Detailed description of the news, including highlights and publisher
characteristics.

Image/Video: Part of the body content that provides a visual illustration to simplify the news
content.

Links: Links to other news sources.

“Fake news” was named word of the year by the Macquarie Dictionary in 2016 [24]. Fake news
has received considerable attention from researchers, with differing definitions from various
view opinions. In [24], the authors defined fake news as “a news article that is intentionally and
verifiably false”. Alcott and Gentzkow [2] provided a narrow definition of fake news as “news
articles that are intentionally and verifiably false, and could mislead readers”. In another
definition, the authors considered fake news as “fabricated information that mimics news
media content in form but not in organizational process or intent” [25]. In [26], the authors
considered fake news in various forms, such as false, misleading, or inventive news, including
several characteristics and attributes of the disseminated information. In [27], the authors
provided a broad definition of fake news as “false news” and a narrow definition of fake news
as “intentionally false news published by a news outlet”. Similar definitions have been employed
in previous fake news detection methods [3], [4], [28], [29].

Characteristics of Fake news:

Although various definitions exist, most fake news has the following common characteristics.
Echo chamber effect: Echo chambers [30] can be broadly defined as environments focusing on
the opinions of users who have the same political leaning or beliefs about a topic. These
opinions are reinforced by repeated interactions with other users with similar tendencies and
attitudes. Social credibility [31] and frequency heuristic [31] (i.e., the trend to search for
information that conforms to preexisting reviews) may be the reason for the appearance of
echo chambers on social networks [24], [32], [33], [34]. When news does not contain enough
information, Social credibility can be used to judge its truthfulness. However, many people still
perceive it as credible and disseminate it, leading to popular acceptance of such news as
credible. A Frequency heuristic forms when people frequently hear the news, leading to natural
approval of the information, even if it is fake news.

Intention to deceive [35]: This characteristic is identified based on the hypothesis that “no one
inadvertently produces inaccurate information in the style of news articles, and the fake news
genre is created deliberately to deceive” [25]. Deception is prompted by political/ideological or
financial reasons [2], [36], [37], [38]. However, fake news may also appear and is spread to
amuse, to entertain, or, as proposed in [39], “to provoke”.
Malicious account: Currently, news on social networks comes from both real people and unreal
people. Although fake news is created and primarily spread by accounts that are not real
people, several real people still spread fake news. Accounts created mainly to spread fake news
are called malicious accounts [27]. Malicious accounts are divided into three main types: social
bots, trolls, and cyborg users [24]. Social bots are social network accounts controlled by
computer algorithms. A social bot is called a malicious account when it is designed primarily to
spread harmful information and plays a large role in creating and spreading fake news [40]. This
malicious account can also automatically post news and interact with other social network
users. Trolls are real people who disrupt online communities to provoke an emotional response
from social media users [24]. Trolls aim to manipulate information to change the views of
others [40] by kindling negative emotions among social network users. Consequently, users
develop strong doubts and distrust them [24]; they will fall into a state of confusion, unable to
determine what is real and what is fake. Gradually, users will doubt the truth and begin to
believe lies and false information. Cyborg users are malicious accounts created by real people;
however, they maintain activities by using programs. Therefore, cyborgs are better at spreading
false news [24].

Authenticity: This characteristic aims to identify whether news is factual [27]. Factual
statements can be proven true or false. Subjective opinions are not considered factual
statements. Only objective opinions are considered factual statements. Factual statements can
never be incorrect. When a statement is published, it is not a factual statement if it can be
disproved [41]. Nonfactual statements are statements that we can agree or disagree with. In
other words, this news is sometimes wrong, sometimes right or completely wrong. Fake news
contains mostly nonfactual statements.

The information is news: This characteristic [27] reflects whether the information is news.

Based on the characteristics of fake news, we provide a new definition of fake news as
follows. “Fake news” is news containing nonfactual statements with malicious accounts that
can cause the echo chamber effect, with the intention to mislead the public.

Concepts related to Fake news: Various concepts regarding fake news exist. Using the
characteristics of fake news, we can redefine these concepts to distinguish them as follows.

False news [42], [43] is news containing nonfactual statements from malicious accounts that
can cause the echo chamber effect with undefined intentions.

Disinformation [44] is news or non-news containing nonfactual statements from malicious

accounts that can cause the echo chamber effect, with the intention to mislead the public.

Cherry-picking [45] is news or non-news containing common factual statements from malicious
accounts and can cause the echo chamber effect, with the intention to mislead the public.
Rumor [46] is news or non-news containing factual or nonfactual statements from malicious
accounts and can cause the echo chamber effect with undefined intentions.

Fake information is news or non-news of nonfactual statements from malicious accounts that
can cause the echo chamber effect, with the intention to mislead the public.

Manipulation [47] is news on markets containing nonfactual statements from malicious

accounts that can cause the echo chamber effect, with the intention to mislead the public.

Deceptive news [2], [24], [27] is news containing nonfactual statements from malicious
accounts that can cause the echo chamber effect, with the intention to mislead the public.

Satire news [48] is news containing factual or nonfactual statements from malicious accounts
that can cause the echo chamber effect, with the intention to entertain the public.

Misinformation [33] is news or non-news containing nonfactual statements from malicious

accounts that can cause the echo chamber effect with undefined intentions.

Clickbait [49] is news or non-news containing factual or nonfactual statements from malicious
accounts that can cause the echo chamber effect, with the intention to mislead the public.

Fake facts [50] are undefined information (news or non-news) comprising nonfactual
statements from malicious accounts that can cause the echo chamber effect, with the intention
to mislead the public.

Propaganda [48] is biased information (news or non-news) comprising undefined statements

(factual or nonfactual) regarding mostly political events from malicious accounts and that can
cause the echo chamber effect, with the intention to mislead the public.

Sloppy journalism [19] is unreliable and unverified information (news or non-news) comprising
undefined statements shared by journalists that can cause the echo chamber effect, with the
intention to mislead the public.

2.2. Fake news detection

2.2.1. What is fake news detection?

Unlike traditional news media, fake news is detected using mainly content-based news
features; for social media, social context-based auxiliary features can aid in detecting fake
news. Thus, in [24], the authors present a formal definition of fake news detection based on the
content-based and context-based features of the news. Given the social
interactions ɛ among n users for news article a, the objective of fake news detection is to
predict whether a is an instance of fake news. This objective is defined by a prediction
function F:ɛ→{0,1} such that,
F(a)={1,0,ifaisapieceoffakenews,otherwise.)
 (1)
Herein, Shu and Sliva define prediction function F as a binary classification function because
fake news detection comprises distorted information from publishers regarding actual news
topics (distortion bias). According to media bias theory [51], a distortion bias is often defined as
a binary classification.
Using the above definition of fake news detection, in this paper, we consider fake news
detection as a multiclassification task. Given a set of n news N={n1,n2,…,nn} and a set
of m labels Ψ, fake news detection identifies a classification function F, such that F:N→Ψ, to
map each news n∈N into the true class with the reliable label in Ψ. Corresponding to the
concepts related to fake news (see Section 2.1) are concepts related to fake news detection,
such as rumor detection and misinformation detection (classification). These concepts are
defined similarly to the fake news detection task.
2.2.2. Fake news detection datasets
In this section, we introduce common datasets that have been recently used for fake news
detection. These datasets were prepared by combining the English datasets presented in
previous papers [19], [52], [53] and enriched by adding missing datasets. In contrast to other
surveys or review papers, we calculated the statistics on 35 datasets, whereas D’Ulizia
et al. [52], Sharma et al. [53], and Khan et al. [19] considered only 27, 23, and 10 datasets,
respectively. Therefore, we list datasets by domain name, type of concept, type of content, and
number of classes. A brief comparison of the fake news datasets is presented in Table 2.
Table 2

A comparison among Fake news datasets.

Type of Type of Number of

Name of dataset Domain
concept content classes
Politics, society,
business, sport, Fake news,
1- ISOT [54], [55] Text 2
crime, technology, real news
health
Text,
2- Fakeddit [56] Society, politics Fake news image, 2, 3, 6
videos
3- LIAR [57] Politics Fake news Text 6
Text,
4- FakeNewsNet [58] Society, politics Fake news 2
image
Text,
5- Stanford Fake Fake news,
Society image, 2
News [59] satire
videos
6- FA-KES [60] Politics Fake news Text 2
 Type of Type of Number of
Name of dataset Domain
concept content classes
Fake news, Text,
7- BREAKING! [61] Society, politics 2, 3
satire image
8- BuzzFeedNews [24] Politics Fake news Text 4
9- FEVER [62] Society Fake news Text 3
10- FakeCovid [63] Health, society Fake news Text 11
11- CredBank [64] Society Rumor Text 2, 5
Fake news,
12- Memetracker [65] Society Text 2
real news
13- BuzzFace [66] Politics, society Fake news Text 4
14- FacebookHoax [67] Science Fake news Text 2
15- Higgs-Twitter [68] Science Fake news Text 2
16- Trust and Believe [69] Politics Fake news Text 2
17- Yelp [70] Technology Fake news Text 2
18- PHEME [71] Society, politics Rumor Text 2
19- Fact checking [72] Politics, society Fake news Text 5
20- EMERGENT [73] Society, technology Rumor Text 3
21- Benjamin Political
Politics Fake news Text 3
News [74]
22- Burfoot Satire Politics,economy,
Satire Text 2
News [75] technology, society
23- MisInfoText [76] Society Fake news Text 5
24- Ott et al.’s Fake
Tourism Text 2
dataset [77] reviews
Politics, society,
25- FNC-1 [78] Fake news Text 4
technology
26-
Politics, society Fake news Text 2
Fake_or_real_news [79]
27- TSHP-17 [80] Politics Fake news Text 2, 6
28- QProp [81] Politics Fake news Text 2, 4
29- NELA-GT-2018[82] Politics Fake news Text 2, 3, 5
30- TW_info [83] Politics Fake news Text 2
Videos,
31- FCV-2018[84] Society Fake news 2
text
 Type of Type of Number of
Name of dataset Domain
concept content classes
Videos,
32- Verification
Society Fake news text, 2
Corpus [85]
image
Politics, society,
33- CNN/Daily Mail [86] business, sport, crime, Fake news Text 4
technology, health
Politics, technology,
34- Tam et al.’s
science, crime, fraud and Rumor Text 5
dataset [87]
scam, fauxtography
35- FakeHealth [88] Health Fake news Text 2
Open in a separate window

Based on the content presented in Table 2, these datasets can be further detailed as follows:

ISOT1 : Both fake news and real news from Reuters; fake news from websites flagged by
PolitiFact and Wikipedia.

Fakeddit: English multimodal fake news dataset including images, comments, and metadata
news.

LIAR2 : English dataset with 12,836 short statements regarding politics collected from online
streaming and two social networks – Twitter and Facebook – from 2007 to 2016.

Stanford Fake News: Fake news and satire stories, including hyperbolic support or
condemnation of a figure, conspiracy theories, racist themes, and discrediting of reliable
sources.

FA-KES: Labeled fake news regarding the Syrian conflict, such as casualties, activities, places,
and event dates.

BREAKING!: English dataset created using the Stanford Fake News dataset and BS detector
dataset3 . The data, including news regarding the 2016 US presidential election, were collected
from web pages.

BuzzFeedNews4 : English dataset with 2283 news articles regarding politics collected from
Facebook from 2016 to 2017.

FakeNewsNet5 : English dataset with 422 news articles regarding society and politics collected
from online streaming and Twitter.
FEVER: English dataset with 185,445 claims regarding society collected from online streaming.

FakeCovid: English dataset with 5182 news articles for COVID-19 health and society crawled
from 92 fact-checking websites, referring to Poynter and Snopes.

CredBank6 : English dataset with 60 million tweets about over 1000 events regarding society
collected from Twitter from October 2014 to February 2015.

Memetracker: English dataset with 90 million documents, 112 million quotes, and 22 million
various phrases regarding society collected from 165 million sites.

BuzzFace: English dataset with 2263 news articles and 1.6 million comments regarding society
and politics collected from Facebook from July 2016 to December 2016. This dataset was
extended in September 2016.

FacebookHoax: English dataset with 15,500 hoaxes regarding science collected from Facebook
from July 2016 to December 2016. Additionally, this dataset identifies posts with over 2.3
million likes.

Higgs-Twitter: English dataset with 985,590 tweets posted by 527,496 users regarding the
science of the new Higgs boson detection collected from Twitter.

Trust and Believe: English dataset with information from 50,000 politician users on Twitter. All
information was labeled manually or using available learning methods.

Yelp: English dataset with 18,912 technology fake reviews collected from online streaming.

PHEME: English and German dataset with 4842 tweets and 330 rumors conversations regarding
society and politics collected from Twitter.

Because of the limited number of manuscript pages, we do not describe further datasets. The
remaining datasets are presented in the Appendix under Description of Datasets.

Based on the above analysis, we compare the criteria of fake news datasets in Fig. 1, followed
by a discussion of observations and the main reason for these observations.
Fig. 1
A comparison among datasets in terms of four criteria.

First, regarding the type of news content, 29 of the 35 datasets contained text data (82.86%);
three of the 35 datasets comprised text, image, and video data (8.57%), namely, Fakeddit,
Stanford Fake News, and Verification Corpus; two of the 35 datasets contained text and image
data (5.71%), namely, FakeNewsNet and Breaking; and only one dataset contained text and
video data (2.86%). No dataset included separate images or videos because previous fake news
detection methods used mainly NLP-based techniques that were highly dependent on text data.
Additionally, labeled image or video data are scarce because annotating them is labor intensive
and costly.

Second, regarding the news domain, 20 and 19 of the 35 datasets focused on society news
(57.14%) and political news (54.29%), respectively, whereas only one dataset contained
economy, fraud/scam, and fauxtography news (2.86%). These findings can be explained by the
fact that fake news is more pertinent and widespread in political and societal domains than in
other domains [89].
Third, regarding the type of fake news concepts, 27 of the 35 datasets contained the fake news
concept (77.14%), followed by rumors (11.43%), satire (8.57%), hoaxes, and real news (5.71%),
and finally, fake reviews (2.86%). Therefore, datasets containing the fake news concept are
generally used for fake news detection applications because fake news contains false
information spread by news outlets for political or financial gains [46].

Finally, regarding the type of applications, the most common application objective of the 35
datasets was fake detection (71.43%), followed by fact-checking (11.43%), veracity
classification, and rumor detection (8.57%) because fake news detection applications can be
used to solve practical problems. Additionally, fake news detection is the most general
application, covering the entire process of classifying false information as true or false. Thus,
fake information datasets are the most relevant for collection [52].
2.2.3. Features of fake news detection

The details of extracting and representing useful categories of features from news content and
context are summarized in Fig. 2.

Fig. 2
Categories of features for fake news detection methods.
Based on the news attributes and discriminative characteristics of fake news, we can extract
different features to build fake news detection models. Currently, fake news detection relies
mainly on news and context information. In this survey, we categorize factors that can aid fake
news detection into seven categories of features: network-, sentiment-, linguistic-, visual-,
post-, user-, and latent-based features.

Linguistic-based features: These are used to capture information regarding the attributes of the
writing style of the news, such as words, phrases, sentences, and paragraphs. Fake news is
created to mislead or entertain the public for financial or political gains. Therefore, based on
the intention of fake news, we can easily extract features related to writing styles that often
appear only in fake news, such as using provocative words to stimulate the reader’s attention
and setting sensational headlines. To best capture linguistic-based features, we divide them
into five common types: lexical, syntactic, semantic, domain-specific, and informality. Lexical
features refer to wording, such as the most salient characters (n-grams) [90], [91], frequency of
negation words, doubt words, abbreviation words, vulgar words [92], and the novelty of
words [93]. Syntactic features capture properties related to the sentence level, such as the
number of punctuations [94], number of function words (nouns, verbs, and adjectives) [93],
frequency of POS tags [95], and sentence complexity [96], [97]. Semantic features capture
properties related to latent content, such as the number of latent topics [98] and contextual
clues [99]. These features are extracted with state-of-the-art NLP techniques, such as
distribution semantics (embedding techniques) and topic modeling (LDA technique) [100].
Domain-specific features capture properties related to domain types in the news, such as
quoted words, frequency of graphs, and external links [101]. Informality features capture
properties related to writing errors, such as the number of typos, swear words, netspeak, and
assent words [27].

Sentiment-based features: This category of features captures properties regarding human

emotions or feelings appearing in the news [102], [103]. These features are identified and
extracted based on the intentions and authenticity characteristics of fake news. They are
classified into two groups: visual polarity and text polarity. The critical factors related to visual
polarity are the number of positive/negative images/videos, number of anxious/angry/sad
images/videos, and number of exclamation marks [27]. These factors capture information
similar to the text polarity.

User-based features: This category of features is identified and extracted based on the
malicious account characteristics of fake news, specifically social bots and cyborg users. User-
based features are properties related to user accounts that create or spread fake news. These
features are classified into two levels, namely, the group level and the individual level [27]. The
individual focuses on exploiting fake or real factors regarding each specific user, such as
registration age, number of followers, and number of opinions posted by users [102], [104].
Meanwhile, the group level focuses on factors regarding the group of users, such as the ratio of
users, the ratio of followers, and the ratio of followees [95], [105].
Post-based features: This category of features is identified and extracted based on the
malicious accounts and news characteristics of fake news. Post-based features are used to
capture properties related to users’ responses or opinions regarding the news shared. These
features are classified into three categories: group, post, and temporal [27]. The post level
focuses on exploiting factors regarding each post [28], such as other users’ opinions regarding
this post (support, deny), main topic, and degree of reliability. The group level focuses on
factors regarding all opinions related to this post [106], such as the ratio of supporting opinions,
ratio of contradicting opinions, and reliability degree [95], [105]. The temporal level notes
factors such as the changing number of posts and followers over time and the sensory
ratio [105].

Network-based features: Network-based features are employed to extract information

regarding the attributes of the media where the news appears and is spread [107]. This
category of features is identified and extracted based on the characteristics of fake news, such
as the echo chamber, malicious account, and intention. Herein, the extractable features are
propagation constructions, diffusion methods, and some factors related to the dissemination of
news, for example, density and clustering coefficient. Therefore, many network patterns can
form, such as occurrence, stance, friendship, and diffusion [24]. The stance network [106] is a
graph with nodes, edges, nodes showing all the text related to the news, and edges between
nodes show similar weights of stances in texts. The co-occurrence network [28] is a graph with
nodes showing users and edges indicating user engagement, such as the number of user
opinions on the same news. The friendship network [105] is a graph with nodes showing users
who have opinions related to the same news and edges showing the followers/followees
constructions of these users. The diffusion network [105] is an extended version of the
friendship network with nodes that indicate users who have opinions on the same news; the
edges show the information diffusion pathways among these users.

Data-driven features: This category of features is identified and extracted based on the data
characteristics of fake news, such as the data domain, data concept, data content, and
application. The data domain exploits domain-specific and cross-domain knowledge in the news
to identify fake news from various domains [108]. The data concept focuses on determining
whether concept drift [109] exists in the news. The data content focuses on considering
properties related to latent content in the news, such as the number of latent topics [98] and
contextual clues [99]. These features are extracted based on state-of-the-art NLP techniques,
such as distribution semantics (embedding techniques) and topic modeling (LDA
technique) [100].

Visual-based features: Few fake news detection methods have been applied to visual
news [24]. This category of features is identified and extracted based on the authenticity, news,
and intended characteristics of fake news. Visual-based features are used to capture properties
related to news containing images, videos, or links [27], [100]. The features in this category are
classified into two groups: visual and statistical. The visual level reflects factors regarding each
video or image, such as clarity, coherence, similarity distribution, diversity, and clustering score.
The statistical level calculates factors regarding all visual content, such as the ratio of images
and the ratio of videos.

Latent features: A critical concept that we need to be aware of herein is latent features that are
not directly observable, including latent textual features and latent visual features. Latent
features are needed to extract and represent latent semantics from the original data more
effectively. This category of features is identified and extracted based on the characteristics of
fake news, such as the echo chamber, authenticity, and news information. Latent textual
features are often extracted by using the news text representation models to create news text
vectors. Text representation models can be divided into three groups: contextualized text
representations, such as BERT [110], ELMo [111], Non-contextualized text representation, such
as Word2Vec [112], FastText [113], GloVe [114], and knowledge graph-based representation,
such as Koloski et al. method [115], RotatE [116], QuatE [117], ComplEx [118]. Contextualized
text representations are word vectors that can capture richer context and semantic
information. Knowledge graph-based representations can enrich various contextual and
noncontextual representations by adding human knowledge representations via connections
between two entities with their relationship based on knowledge graphs. News text
representations can be not only used as inputs for traditional machine learning
models [119] but also integrated into deep learning models, such as neural networks [115],
recurrent networks [120], and transformers [110], [121], [122], and GNNs-based
models [123], [124], [125] for fake news detection. Latent visual features are often extracted
from visual news, such as images and videos. Latent visual features are extracted by using
neural networks [126] to create a latent visual representation containing an image pixel tensor
or matrix.
2.2.4. Fake news detection techniques

Fig. 3 shows an overview of fake news detection techniques. Previous related

papers [21], [27], [42], [46], [53], [79], [107] demonstrated that fake news detection techniques
are often classified into four categories of approaches: content-based approaches, including
knowledge-based and style-based approaches, context-based approaches, propagation-based
approaches, multilabel learning-based approaches, and hybrid-based fake news detection
approaches. Let Ψa be one of the corresponding output classes of the fake news detection task.
For example, Ψa∈ {real, false} or Ψa∈ {nonrumor, unverified rumor, false rumor, true rumor}
or Ψa∈ {true, false}.
Fig. 3
Categories of fake news detection.
Knowledge-based detection:
Given news item a with a set of knowledge denoted by triple K=(S,P,O) [127], where S={s1,s2,
…,sk} is a set of subjects extracted from news item a, P={p1,p2,…,pk} is a set of predicates
extracted from news item a, O={o1,o2,…,ok} is a set of objects extracted from news item a.
Thus, kai=(si,pi,oi) ∈K, 1≤i≤n, is called a knowledge. For example: we have a news as “John
Smith is a famous doctor at a central hospital”; from this statement, we
have kai=(JohnSmith,Profession,Doctor). Assume that we have a set of true
knowledge Kt=(St,Pt,Ot), where ktal=(stl,ptl,otl) ∈Kt, 1≤l≤m. Let GK be a true knowledge graph
including a set of true knowledge, where nodes represent a set of (St,Ot)∈Kt and edges
represent a set of (Pt)∈Kt, the aim of a knowledge-based fake news detection method is to
define a function F to compare kai=(si,pi,oi)∈K with ktal=(stl,ptl,otl) ∈Kt, such
that: F:kaiGK−→−Ψai. Function F is used to assign a label Ψai∈[0,1] to each triple (si,pi,oi) by
comparing it with all triples (stl,ptl,otl) on graph GK, where labels 0 and 1 indicate fake and real,
respectively. Function F can be defined as F(kai,GK)=Pr(edgepiisalinkfromsˆitooˆionGK),
where Pr is the probability; sˆi and oˆi are the matched nodes to si and oi on GK,
respectively. sˆi and oˆi are identified as sˆi=argminstl|J(si,stl)|<ξ and oˆi=argminotl|J(oi,otl)|<ξ,
respectively, where ξ is a certain threshold; J(si,stl) is a function to calculate the distance
between si and sti and it is the similar for J(oi,otl). For example, when |J(si,stl)|=0 or |J(si,stl)|
<ξ, we can regard si as the same as sti. The techniques in this category are proposed based on
the authenticity and news characteristics of fake news. The objective of knowledge-based
techniques is to employ external sources to fact-check news statements. The fact-checking step
aims to identify the truth of a statement corresponding to a specific context [72]. It can be
implemented automatically (computational-oriented [128]) or manually (expert-
based [101], [129], [130], crowd-sourced [67], [131]).

Style-based detection:
Given a news item a with a set of fas style features, where fas is a set of features regarding the
news content. Style-based fake news detection is defined as binary classification to identify
whether news item a is fake or real, which means that we have to find a mapping
function F such that F:fas→Ψa. The techniques in this category are proposed based on the
intention and news characteristics of fake news. The objective of style-based techniques is to
capture the distinct writing style of fake news. Fake news employs distinct styles to attract the
attention of many people and stand out from ordinary news. The capturing step of the writing
styles was built automatically. However, two techniques must be observed as criteria: style
representation techniques [132], [133], [134] and style classification
techniques [28], [91], [135].

Context-based detection:
Given news item a with a set of fac context features, where fac includes news text, news
source, news publisher, and news interaction. Context-based fake news detection is defined as
the task of binary classification to identify whether news item a is fake or real, which means
that we have to find a mapping function F such that F:fac→Ψa. The techniques in this category
are proposed based on the malicious account and news characteristics of fake news. The
objective of source-based techniques is to capture the credibility of sources that appear,
publish, and spread the news [27]. Credibility refers to people’s emotional response to the
quality and believability of news. The techniques in this category are often classified into two
approaches: (i) assessing the reliability of sources where the news appeared and is spread
based on news authors and publishers [136], [137] and (ii) assessing the reliability of sources
where the news appeared and is spread based on social media users [105], [138], [139].

Propagation-based detection:
Given news item a with a set of fap propagation patterns features for news. Propagation-based
fake news detection is defined as binary classification to identify whether news item a is fake or
real, which means that we have to develop a mapping function F such that F:fap→Ψa. The
techniques in this category are proposed based on the echo chamber effect and news
characteristics of fake news. The objective of propagation-based techniques is to capture and
extract information regarding the spread of fake news. That is, the methods in this category aim
to detect fake news based on how people share it. These techniques are often grouped into
two small categories:
(i) using news cascades [140], [141] and
(ii) using self-defined propagation graphs [142], [143], [144], [145].
Multilabel learning-based detection:
Let χ∈Rd be the d-dimension input feature matrix; hence, news item a=[a1,…,ad]∈χ; and
let Γ={real,fake}l be the label matrix, such that Ψ=[Ψ1,…,Ψl]∈Γ, where l is the number of class
labels. Given a training set {(a,Ψ)}, the task of multilabel learning detection is to learn a
function F:χ→Γ to predict Ψˆ=F(a). Multilabel learning-based detection is a learning method
where each news item in the training set is associated with a set of labels. The techniques in
this category are proposed based on the echo chamber effect and news characteristics of fake
news. The objective of multilabel learning-based techniques is to capture and extract
information regarding the news content and the news latent text. The techniques in this
category are often classified into four approaches: (i) using style-based
representation [17], [115], [146], [147]; (ii) using style-based
classification [15], [29], [148], [149], [150], [151]; (iii) using news cascades [140], [152]; and (iv)
using self-defined propagation graphs [4], [16], [125], [153].

Hybrid-based detection: This method is a state-of-the-art approach for fake news detection
that simultaneously combines two previous approaches, such as content-context [154], [155],
propagation-content [147], [156], and context-propagation [4], [14]. These hybrid methods are
currently of interest because they can capture more meaningful information related to fake
news. Thus, they can improve the performance of fake news detection models.

A critical issue that needs to be discussed is fake news early detection. Early detection of fake
news provides an early alert of fake news by extracting only the limited social context with a
suitable time delay compared with the appearance of the original news item. Knowledge-based
methods are slightly unsuitable for fake news early detection because these methods depend
strongly on knowledge graphs; meanwhile, newly disseminated news often generates new
information and contains knowledge that has not appeared in knowledge graphs. Style-based
methods can be used for fake news early detection because they depend mainly on the news
content that allows us to detect fake news immediately after news appears and has not been
spread. However, style-based fake news early detection methods are only suitable for a brief
period because they rely heavily on the writing style, which creators and spreaders can change.
Propagation-based methods are unsuitable for fake news early detection because news that is
not yet been disseminated often contains very little information about its spread. To the best of
our knowledge, context-based methods are most suitable for fake news early detection
because they depend mainly on the news surroundings, such as news sources, news publishers,
and news interactions. This feature allows us to detect fake news immediately after news
appears and has not been spread by using website spam detection [157], distrust link
pruning [158], and user behavior analysis [159] methods. In general, early detection of fake
news is only suitable for a brief period because human intelligence is limitless. When an early
detection method of fake news is applied, it will not be long until humans create an effective
way to combat it. This issue is still a major challenge for the fake news detection field.
2.3. Understanding graph neural networks
In this section, we provide the background and definition of a GNN. The techniques, challenges,
and types of GNNs are discussed in the following section. Before presenting the content of this
section, we introduce the notations used in this paper in Table 3.
Table 3

Descriptions of notations.

Notations Descriptions
|.| The length of a set
G A graph
V The set of nodes in a graph
v A node in a graph
E The set of edges in a graph
eij An edge between two nodes vi,vj in a graph
A The graph adjacency matrix
D The degree matrix of A. Dii=∑nj=1Aij
n The number of nodes
m The number of edges
r The set of relations of edges
d The dimension of node feature vector
c The dimension of edge feature vector
xevi,vj∈Rc The feature vector of edge eij
xnv∈Rd The feature vector of node v
Xe∈Rm×c The edge feature matrix of a graph
X∈Rn×d The node feature matrix of a graph
X(t)∈Rn×d The node feature matrix at the time step t
Open in a separate window
2.3.1. What is a graph?
Before we discuss deep learning models on graph structures, we provide a more formal
description of a graph structure. Formally, a simple graph is presented as G={V,E},
where V={v1,v2,…,vn} is the set of nodes, and E={e11,e12,…,enn} is the set of edges
where eij=(vi,vj) ∈E, 1≤i,j≤n. In which, vi and vj are two adjacent nodes. The adjacency
matrix A is a n × n matrix with
Aij={1,0,ifeij∈E,ifeij∉E.)
 (2)

We can create improved graphs with more information from simple graphs, such as attributed
graphs [6], multi-relational graphs [160].

Attributed graphs are the extended version of simple graphs. They are obtained by adding the
node attributes X or the edge attributes Xe, where X∈Rn×d is a node feature matrix
with xnv∈Rd indicating the feature vector of a node v; Xe∈Rm×c is an edge feature matrix
with xevi,vj∈Rc indicating the feature vector of an edge eij.
Spatial–Temporal graphs are special cases of attributed graphs, where the node attributes
automatically change over time. Therefore, let X(t) be a feature matrix of the node
representations at tth time step, a spatial–temporal graph is defined as G(t)= {V,E,X(t)},
where X(t)∈Rn×d.
Multi-relational graphs are another extension version of simple graphs that include edges with
different types of relations τ. In these cases, we have eij=(vi,vj)∈E →eij=(vi,τ,vj)∈E. Each edge
has one relation adjacency matrix Aτ. The entire graph can be created an adjacency
tensor A ∈Rn×r×n. The multi-relational graphs can be divided into two subtypes: heterogeneous
and multiplex graphs.
Heterogeneous graphs: Here, nodes can be divided into different types. That
means V=V1∪V2∪...∪Vk, where for i≠j, Vi∩Vj=∅. Meanwhile, edges must generally satisfy the
conditions following the node types. Then, we have eij=(vi,τ,vj)∈E→eij=(vi,τh,vj)∈E,
where vi∈Vt,vj∈Vk and t≠k.
Multiplex graphs: Here, graphs are divided into a set of k layers, where each node belongs to
one layer, and each layer has a unique relation called the intralayer edge type. Another edge
type is the interlayer edge type. The interlayer connects the same node across the layers. That
means G={Gi,i∈{1,2,…,k}},Gi={Vi,Ei}, with Vi={v1,v2,
…,vn}, Ei=Eintrai∪Einteri, Eintrai={elj=(vl,vj),vl,vj∈Vi}, Einteri={elj=(vl,vj),vl∈Vi,vj∈Vh,1≤h≤k,h≠i}.
2.3.2. What are graph neural networks?
GNNs are created using deep learning models over graph structure data, which means deep
learning models deal with Euclidean space data; in contrast, GNNs [6], [161], [162], [163] deal
with non-Euclidean domains. Assume that we have a graph G={V,E} with adjacency
matrix A and node feature matrix (or edge feature matrix) X (or Xe). Given A and X as inputs,
the main objective of a GNN is to find the output, i.e., node embeddings and node classification,
after the k-th layer is: H(k)=F(A,H(k−1);θ(k)), where F is a propagation function; θ is the
parameter of function F, and when k = 1, then H(0)=X. The propagation function has a number
of forms. Let σ(⋅) be a non-linear activation function, e.g., ReLU; W(k) is the weight matrix for
layer k; Aˆ is the normalized adjacency matrix and calculated as Aˆ=D−0.5A⊤D−0.5 with D, is the
diagonal degree matrix of A⊤, that is calculated as Dii=∑jA⊤ij; A⊤=A+I with I is the identity
matrix. A simple form of the propagation function is often used: F(A,H(k))=σ(AH(k−1)W(k)). In
addition, the propagation function can be improved to be suitable for GNN tasks as follows:
For the node classification task, function F often takes the following form [164]:
F(A,H(k))=σ(AˆH(k−1)W(k))
 (3)
For the node embeddings task, function F often takes the following form [165]:
F(A,H(k))=σ((Qϕ(H(k−1)eMe)Q⊤⊙Aˆ)H(k−1)W(k))
(4)
where Q is a transformer representing whether edge e is connected to the given node
and Q⊤=T+I; Me the learnable matrix for the edges; ϕ is the diagonalization operator; ⊙ is the
element-wise product; H(k−1)e is the hidden feature matrix of edges in the (k−1)-th layer,
where H0e=Xe (Xe is the edge feature matrix). The Qϕ(H(k−1)eMe)Q⊤ is to normalize the
feature matrix of edges. The Qϕ(H(k−1)eMe)Q⊤⊙Aˆ is to fuse the adjacency matrix by adding
the information from edges.

More choices of the propagation function in GNNs are detail presented in Refs. [13], [165].
Early neural networks were applied to acyclic graphs by Sperduti et al. [166] in 1997. In 2005,
Gori et al. [167] introduced the notion of GNNs, which were further detailed by Scarselli
et al. [168] in 2009 and by Gallicchio et al. [169] in 2010. According to Wu et al. [6], GNNs can
be divided into four main taxonomies: conventional GNNs, graph convolutional networks, graph
autoencoders, and spatial–temporal graph neural networks. In the next subsections, we
introduce the categories of GNNs.

Conventional graph neural networks

(GNNs∗) which are an extension of recurrent neural networks (RNNs), were first introduced by
Scarselli et al. [168] by considering an information diffusion mechanism, where states of nodes
are updated and information is exchanged until a stable equilibrium is obtained [167], [168]. In
these GNNs, the function F is also defined as Eq. (3). However, the feature matrix of the k-th
layer H(k) is updated using different equation as follows:
H(k)vj=∑vi∈N(vj)F(xnvj,xe(vi,vj),xnvi,H(k−1)vi)
(5)
where N(vj) is the set of neighbor nodes of node vj, F is a parametric function, H(k)vj is the
feature vector of node vj for the k-th layer, and H(0)vj is a random vector.

Graph convolutional networks (GCNs) were first introduced by Kipf and Welling [164]. They are
capable of representing graphs and show outstanding performance in various tasks. In these
GNNs, after the graph is constructed, the function F is also defined as Eq. (3). However, the
recursive propagation step of a GCN at the k-th convolution layer is given by:
H(1)=σ(AˆH(0)W(1)+b(1))
(6)

Hence,
 H(2)=σ(AˆH(1)W(2)+b(2))
(7)

That means:

H(k)=σ(AˆH(k−1)W(k)+b(k))
(8)
where H(0)=X. σ(⋅) is an activation function. W(k)∈Rm×d, k={1,2,3,…} is a transition matrix
created for the k-th layer. b(1) and b(2) are the biases of two layers.
Graph autoencoders (GAEs) are deep neural architectures with two components: (i) the
encoder, which converts nodes on the graph into a vector space of latent features, and (ii) the
decoder, which decodes the information on the graph from the latent feature vectors. The first
version of GAEs was introduced by Kipf and Welling [170], [171]. In these GNNs, the form of
function F is redefined as the following Equation:
F(A˜,H(k))=σ(A˜H(k−1)W(k))
(9)
where A˜=φ(ZZ⊤) is the reconstructed adjacency matrix and φ is the activation function of the
decoder composition. Z is the output of the encoder composition. In these GAEs, the GCNs are
used in the encoder step to create the embedding matrix; therefore, Z is calculated based on
Eq. (3). Thus, Z=F(Aˆ,H(k)) with F(⋅) corresponds to the case of GCNs. Z⊤ is the transpose matrix
of Z.

Spatial–temporal graph neural networks (STGNNs) in various real-world tasks are dynamic as
both graph structures and graph inputs. To represent these types of data, a spatial–temporal
graph is constructed as introduced in Section 2.3.1. However, to capture the dynamicity of
these graphs, STGNNs have been proposed for modeling the inputs containing nodes with
dynamic and interdependency. STGNNs can be divided into two approaches: RNN-based and
CNN-based methods.
For the RNN-based approach, to capture the spatial–temporal relation, the hidden states of
STGNNs are passed to a recurrent unit based on graph convolutions [172], [173], [174]. The
propagation function form of STGNNs is also shown in Eq. (3). However, the value of the k-th
layer is calculated as follows:
H(t)=σ(WXn(t)+UH(t−1)+b)
(10)
where Xn(t) is the node feature matrix at time step t. After using graph convolutions, Eq. (10) is
recalculated as follows:
H(t)=σ(GCN(Xn(t),Aˆ;W)+GCN(H(t−1),Aˆ;U)+b)
(11)
where GCN is one of GCNs model. U∈Rn×n is the eigenvector matrix ranked by eigenvalues
with U⊤U=I.

For the CNN-based approach, RNN-based approaches recursively handle spatial–temporal

graphs. Thus, they must iterate the propagation process and therefore they have limitations
regarding the propagation time and gradient explosion or vanishing
problems [175], [176], [177]. CNN-based approaches can solve these problems by exploiting
parallel computing to achieve stable gradients and low memory.

Attention-based graph neural networks (AGNNs) [178] remove all intermediate fully
connected layers and replace the propagation layers with an attention mechanism that
maintains the structure of the graph [179]. The attention mechanism allows learning a dynamic
and adaptive local summary of the neighborhoods to obtain more accurate predictions [180].
The propagation function form of the AGNN is shown in Eq. (3). However, the AGNN includes
graph attention layers. In each layer, a shared, learnable linear transformation M∈Rth×dh,
where h is the number of the t-th hidden layer, dh is the dimensional of the t-th hidden layer, is
used for the input features of every node as follows:
H(t)=σ(M(t)H(t−1))
(12)
where the row-vector of node vi defined as follows:
H(t)vi=∑vj∈N(vi)∪{i}M(t−1)ijH(t−1)j
(13)

where

M(t−1)ij=φ([β(t−1)cos(H(t−1)i,H(t−1)j)]vj∈N(vi)∪{i})
(14)
where β(t−1)∈R is an attention-guided parameter of propagation layers. Note that the value
of β of propagation layers is changed over hidden states. φ(⋅) is the activation function of
propagation layer.

3. Survey methodology

In this study, we conducted a systematic review of fake news detection articles using GNN
methods, including three primary steps: “literature search,” “selection of eligible papers,” and
“analyzing and discussing” [181]. The research methodology is illustrated in Fig. 4:
Fig. 4
Flow diagram of research methodology.

The literature search is used to select peer-reviewed and English-language scientific papers
containing the following keywords: “GNN” OR “graph neural network” OR “GCN” OR “graph
convolutional network” OR “GAE” OR “graph autoencoder” OR “AGNN” OR “attention-based
graph neural network” combined with “fake news” OR “false news” OR ”rumour” OR ”rumor”
OR “hoax” OR “clickbait” OR “satire” OR “misinformation” combined with “detection”. These
keywords were extracted from Google Scholar, Scopus, and DBLP from January 2019 to the end
of Q2 2021.

The selection of eligible papers is used to exclude the nonexplicit papers on fake news
detection using GNNs. To select the explicit papers, we specify a set of exclusion/inclusion
criteria. The inclusion criteria were as follows: written in English, published after 2019, peer-
reviewed, and retrieved full-text. The exclusion criteria were as follows: papers of reviews,
surveys, and comparisons or only presented mathematical models.

Analysis and discussion papers are used to compare the surveyed literature and capture the
main challenges and interesting open issues that aim to provide various unique future
orientations for fake news detection.

By the above strategy, a final total of 27 papers (5 papers in 2019, 16 papers in 2020, and 6
papers for the first 6 months of 2021) are selected for a comprehensive comparison and
analysis. These selected papers are classified into four groups based on GNN taxonomies (see
Section 2.3.2), including conventional GNN-based, GCN-based, AGNN-based, and GAE-based
methods. In the next step, eligible papers are analyzed via the criteria of the method’s name,
critical idea, loss function, advantage, and disadvantage.

4. Quantitative analysis of eligible papers

Previous fake news detection approaches have mainly used machine

learning [74], [92], [182], [183], [184] and deep
learning [95], [120], [185], [186], [187], [188], [189] for classifying news as fake or real, rumor or
not a rumor, and spam or not spam. Various surveys and review papers regarding fake news
detection using machine learning and deep learning have been published. In this paper, we
discuss in detail the most current GNN-based fake news detection approaches. Using the
research methodology in Section 3, a final total of 27 papers published after 2019 using GNNs
for fake news detection were selected for a more detailed review in the following
subsections. Table 4 presents comparisons among previous studies in terms of model name,
referral code (Table 5), authors, year of publication, type of GNN, datasets, performance, and
approach-based fake news detection.
Table 4

Comparison of surveyed methods using GNNs for fake news detection.

Method’s PY and Approach-

Authors Dataset Performance
name TG based
2019,
1-Monti et al. [3] Tweets ROC AUC:92.7% Propagation
GCN
Method’s PY and Approach-
Authors Dataset Performance
name TG based
2019,
2- !GAS [123] Spam dataset F1: 82.17% Context
GCN
2019,
3-MGCN [190] Liar Acc: 49.2% Content
GCN
4-!Chang Li 2019, Acc: 67.03–
[191] 10,385 news articles Context
et al. GCN 88.89%
2019, Acc: 70.45–
5-Benamira AGNN 84.25%
[192] Horne et al. [74] Content
et al. 2 2019, Acc: 72.04–
GCN 84.94%

6-Marion 2020,
[153] FakeNewsNet Acc: 73.3% Propagation
et al. 1 GCN

FakeNewsNet: Acc: 79.2–

2020, Propagation,
7-Yi Han et al. [4] Politifact 80.3%
GNN∗ context
GossipCop Acc: 82.5–
83.3%
2020,
8-FakeNews [193] Covid-19 tweets ROC: 95% Content
GNN∗
2020, Twitter 15 [140], Acc: 87.67% Context,
9-GCAN3 [14]
GCN Twitter 16 [140] Acc: 90.84% propagation
Task 1: MCC:
10-Nguyen 2020, 36.1–41.9%
[155] Tweets Content
et al. GCN Task 2: MCC:
−8.1–1.51%
GCN: T-MCC:
2%
11-Pehlivan 2020, DGCNN: T-
[194] Covid-19 Tweets Content
et al. 4 GCN MCC: 2.3%
M-FCN: T-MCC:
3.5%
Weibo [99], Acc: 96.8%
2020,
12-*Bi-GCN [16] Twitter 15 [140], Acc: 88.6% Propagation
GCN
Twitter 16[140] Acc: 88.0%
13-VGCN- 2020, 1600 images with
[195] F1: 84.37% Content
ItalianBERT5 GCN metadata
Method’s PY and Approach-
Authors Dataset Performance
name TG based
Karate [197],
Dolphin [198], Improve the
2020,
14-*GCNSI [196] Power grid [199], best method Propagation
GCN
Jazz [200], by about 15%
Ego-Facebook
F1 above
2020, FakeNewsNet, 92.97%
15-SAFER6 [15] Context
GNN∗ FakeHealth F1 above
58.34%
16-! 2020, Acc: 79.2–
[201] Twitter [202], [203] Propagation
GCNwithMRF GCN 83.9%
Acc: 93.4–
Weibo [99], 94.4%
7 2020,
17-*Lin et al. [124] Twitter 15[140] Acc: 84–85.6% Propagation
GAE
Twitter 16[140] Acc: 85.2–
88.1%
18-*Malhotra 2020, Twitter 15 [99] Acc: 86.6% Propagation,
[147]
et al. GCN Twitter 16 [140] Acc: 86.5% content
Comments on
2020, Acc: 71.09%
19-!FauxWard [149] Twitter Content
GCN Acc: 75.36%
Comments on Reddit
Weibo [99], Acc: 95.0%
2020,
20- KZWANG8 [156] Twitter 15[140], Acc: 91.1% Propagation
GCN
Twitter 16[140] Acc: 90.7%
2020, FakeNewsNet, Context,
21- FANG9 [5] AUC: 75.18%
GNN∗ PHEME content

Acc: 69.0–
2021, Twitter [140] 77.0%
22-*GraphSAGE [125] Propagation
GCN PHEME Acc: 82.6–
84.2%
MCC: 33.12–
23- Bert-GCN 2021, Covid-19 and 5G 47.95%
[150] Content
Bert-VGCN GCN tweets MCC: 39.10–
49.75%
24-*Lotfi [204] 2021, PHEME F1: 80% Content,
GCN (rumor) propagation
F1: 79% (non-
Method’s PY and Approach-
Authors Dataset Performance
name TG based
rumor)
Acc: 79.2–
2021, Twitter 15[140],
25-*SAGNN [151] 85.7% Content
GCN Twitter 16[140]
Acc: 72.6-86.9%
2021, Fact-checking, Acc: 61.55% Content,
26-AA-HGNN [17]
AGNN BuzzFeedNews Acc: 73.51% context
2021, Acc: 63.8–
27-*EGCN [154] PHEME Propagation
GCN 84.1%
Open in a separate window
PY and TG: Publication year and Type of GNNs. MCC: Matthews correlation coefficient. T-MCC:
MCC for test dataset. M-FCN: MALSTM-FCN model.

Table 5

Code source.

Refe
Code source
r
1
[Link]
2
[Link]
3
[Link]
4
[Link]
5
[Link]
6
[Link]
7
[Link]
8
[Link]
9
[Link]
Open in a separate window

Using the relationships among the information in Table 4, we compare quantitatively surveyed
methods in terms of four distribution criteria of GNN-based fake news detection approaches, as
shown in Fig. 5.
Fig. 5
A comparison of four distribution criteria of GNN-based fake news detection approaches.

The number of surveyed papers from 2019 to 2021 (the end of Q2) regarding fake news
detection using GNNs shows that this problem is attracting increasing attention from system
practitioners (increasing 40.74% from 2019 to 2020). Although in 2021, only 22.22% of articles
on fake news detection focused on using GNNs, Q2 has not yet ended, and we believe that the
last two quarters of the year will produce more articles in this field, considering the outbreak of
fake news related to COVID-19 and the challenges of this problem.

With regard to the type of news concepts employed (types of objectives), 14 of the 27 surveyed
papers are related to fake news detection (51.85%), followed by rumors and spam detection
(29.63%, 7.41%), whereas other types of detection constitute only 3.7%. A likely reason for
these results is the creation and spread of fake news correspond to active economic and
political interests. That is, if fake news is not detected and prevented in a timely manner,
people will suffer many deleterious effects. Additionally, as analyzed above, an equally
important reason is that datasets used for fake news detection are now richer and more fully
labeled than other datasets (see Section 2.2.2).
With regard to GNN-based techniques, the authors predominantly (74.07%) used GCNs for fake
news detection models, followed by GNN-based methods (14.81%), GAE, and AGNN (3.7%).
This choice is attributable to the suitability of GCNs for graph representations in addition to
having achieved state-of-the-art performance in a wide range of tasks and applications [13].

Finally, one-third of the propagation-based and content-based approaches (33.33%) were

published, followed by hybrid-based (22.22%) and context-based (11.11%) approaches. This
result is attributable to propagation-based and context-based approaches using mainly news
information on network structures, users, and linguistics simultaneously. This information is
most consequential for fake news detection.

5. Literature survey

In this section, we survey papers using graph neural networks for fake news detection. Based
on GNN taxonomies (see Section 2.3.2), we categorized GNN-based fake news detection
methods into conventional GNN-based, GCN-based, AGNN-based, and GAE-based methods, as
shown in Table 6.
Table 6

GNN-based detection methods categorization.

Category Publication
Conventio
[4], [5], [15], [192], [193]
nal GNN
[3], [14], [16], [123], [125], [147], [149], [150], [151], [153], [154], [155], [156], [1
GCN
94], [201], [204]
AGNN [17], [192]
GAE [124]

Conventional GNN-based methods (GNN∗) are pioneering GNN-based fake news detection
methods. These methods apply a similar set of recurrent parameters to all nodes in a graph to
create node representations with better and higher levels.

GCN-based methods (GCN) often use the convolutional operation to create node
representations of a graph. Unlike the conventional GNN-based approach, GCN-based methods
allow integrating multiple convolutional layers to improve the quality of node representations.

AGNN-based methods are constructed mainly by feeding the attention mechanism into graphs.
Thus, AGNNs are used to effectively capture and aggregate significant neighbors to represent
nodes in the graph.
GAE-based methods are unsupervised learning approaches to encode nodes on a graph into a
latent vector and decode the encoded information to reconstruct the graph data to create node
representations by integrating latent information.

Most approaches proposed in the surveyed papers for detecting false information are used to
solve a classification problem task that involves associating labels such as rumor or nonrumor
and true or false with a particular piece of text. In using GNNs for fake news detection,
researchers have employed mainly conventional GNNs and GCNs to achieve state-of-the-art
results. On the other hand, some researchers have applied other approaches, such as GAE and
AGNN, to predict their conforming labels.

5.1. Detection approach based on GNNs∗

GNNs∗ represent the first version of GNNs and improved the performance of fake news
detection of machine learning and deep learning methods that use non-Euclidean data.
Han et al. [4] exploited the capability of GNNs∗ using non-Euclidean data to detect the
difference between news propagation methods on social networks. They then classified the
news into two labels of fake and real news by training two instances of GNNs∗. In the first case,
GNNs∗ were trained on complete data. The second case involved training GNNs∗ using partial
data. In the second case, unlike conventional GNNs, two techniques – gradient episodic
memory and elastic weight consolidation – were used to build GNNs∗ with continual learning
aimed at the early detection of fake propagation patterns. This method can obtain superior
performance without considering any text information compared with state-of-the-art models.
In particular, time and cost are saved as the dataset grows when training the new data because
the entire dataset is not retrained. However, one major limitation is that the strong forgetting
occurrence is not solved by extracting more features, including the “universal” features. Hamid
et al. [193] introduced a method to detect malicious users who spread misinformation by
analyzing tweets related to conspiracy theories between COVID-19 and 5G networks. This
method includes two substrategies: (i) content-based fake news detection and (ii) context-
based fake news detection. The second strategy is implemented based on GNNs∗ to train the
GNNs∗ representation and to classify 5G networking comments into three categories:
nonconspiracy, conspiracy, and other conspiracies. The obtained performance in terms of
average ROC is quite good (0.95%) because it captures mostly information related to the textual
and structural aspects of news. However, neither textual nor structural information was used
simultaneously. Nguyen et al. [5] proposed a model named FANG for fake news detection
based on the news context by considering the following steps: (i) extracting features regarding
the news, such as the source, users, and their interactions, and posting a timeline; (ii)
constructing two subhomogeneous graphs, namely, news source and user; and (iii) using an
unsupervised model over two subgraphs separately to model neighbor relations. Moreover, the
authors used pretrained detection networks to detect news content as extension information.
FANG can capture a news context with higher fidelity than recent graphical and nongraphical
models. In particular, FANG still achieves robustness even with limited training data. However,
features such as users and their interactions are extracted before being fed into the FANG.
Therefore, some errors regarding textual encoding and emotion detection can occur, and they
are provided to FANG. Another limitation is the rapid obsolescence of contextual datasets
because we cannot retrieve hyperlinks and other traces at the query time, as they might no
longer be available.
Unlike other contemporary work on GNN∗, the fake news detection task is introduced above.
Chandra et al. [15] presented a method called SAFER with three distinct features. (i) They
constructed a GNN∗ model with the same heterogeneous input graph for two types of edges
and nodes. (ii) They determined context features by exploiting the impact of online social
communities without using user profiles. (iii) They only used the network information of online
users to evaluate these communities’ roles, but their results were still better than those of
previous approaches. The authors proposed a relational GNN∗ and a hyperbolic GNN∗ to model
user and community relations. The relational GNN obtained better results than conventional
GNNs. However, the results of the hyperbolic GNN∗ were comparable only to the other GNN∗.
Therefore, modeling users/communities for truly hierarchical social network datasets is a
challenge that needs to be addressed in the future.

5.2. Detection approach based on GCNs

The GCN-based approach is a category of methods that are used mostly for fake news detection
and rely on GNNs. GCNs are an extension of GNNs that derive the graph structure and integrate
node information from neighborhoods based on a convolutional function. GCNs can represent
graphs and achieve state-of-the-art performance on various tasks, including fake news
detection.
Lu et al. [14] presented a novel method for fake news detection of tweets called GCAN, which
includes five main steps as follows: (i) extract quantified features related to users; (ii) convert
words in news tweets into vectors; (iii) represent aware propagation methods of tweets among
users; (iv) capture the correlation between tweet context and user interactions and between
tweet context and user propagation; and (v) classify tweets as fake or real news by combining
all learned representations. GCAN exhibits outstanding performance with reasonable
explainability. The main contribution of this study is the integration of dual coattention
mechanisms with GCNs. The first mechanism simultaneously captures the relations between
the tweet context and user interactions. The second mechanism simultaneously captures the
relations between the tweet context and user propagation. This method can be considered an
enriching version of GCNs. The form of GCAN was improved from GCNs as
follows: Hs=tanh(WsS+(WgG)F⊤), Hg=tanh(WgG+(WsS)F), where S represents the embeddings
of the relations between the tweet context and user interactions; G represents the embeddings
of the relations between tweet context and user propagation; Ws and Wg represent matrices of
learnable parameters; F and F⊤ are the transformation matrix and its transpose, respectively.
Based on the inherent aggregation mechanism of the GNN, Zhang et al. [151] proposed a
simplified GNN method called SAGNN to calculate the degree of interaction between Twitter
users and other users for rumor detection. Unlike the convolution layer of the conventional
GCN, the SAGNN does not contain the weight matrix W. Moreover, the identification of the
adjacency matrix was different from that of conventional GCNs. Thus, the two layers in the
SAGNN are defined as H(1)=σ(H(0)E), where H(1) is called the embedding layer and E is the
word embedding matrix. Hence, H(2)=σ(A˜H(1)), where H(2) is called the aggregation layer;
and A˜=I+uB+vC, where u,v are learnable parameters of SAGNN. Matrix B is calculated as
follows: ifviisthe parentofvj thenBij=1otherwise0; whereas matrix C is defined
as ifviis achildofvj thenCij=1otherwise0.

Ke et al. [156] constructed a heterogeneous graph, namely, KZWANG, for rumor detection by
capturing the local and global relationships on Weibo between sources, reposts, and users. This
method comprises three main steps as follows: (i) word embeddings convert text content of
news into vectors using a multihead attention mechanism,

T=MultiHead(Q,K,V)=Concat(head1,…,headh)Wo
(15)
where headi=attention(QWQi,KWKi,VWVi) with
Q∈Rnq×d,K∈Rnk×d,V∈Rnv×d are sentences of query, key, and value; nq,nk,nv are the number
of words in each sentence; and attention(Q,K,V ) =Softmax(QK⊤dk√)V; (ii) propagation and
interaction representations are learned via GCNs; and (iii) graph construction builds a model of
potential interactions among users: P=H(k)=σ(AˆH(k−1)W(k−1)). The difference between this
model and conventional GCNs is that KZWANG is a combination of the news text representation
using a multihead attention mechanism and propagation representation using GCNs. Thus, the
outputs of the GCN layer and the multihead attention layer are the inputs of rumor
classification: R = Softmax(TP + b), where T is the text representation matrix. P is the
propagation representation matrix. R is the output of the whole model.
Lotfi et al. [204] introduced a model that includes two GCNs: (i) a GCN of tweets, such as source
and reply as T=H(k)=σ(AˆTH(k−1)W(k−1)); (ii) GCN of users, such as interaction among users
as Re=H(k)=σ(AˆReH(k−1)W(k−1)) where AT is the adjacency matrix of the GCN of tweets and
determined as if(tweetirepliestotweetj)or(i=j) thenAijT=1otherwise0. Meanwhile, ARe is the
adjacency matrix of the GCN of users and defined as
follows: if(userisentmtweetstouseriinconversation)or(i=j) thenAijRe=1 otherwise0.
And H(0)=X is determined
as:if(thereishighfrequencywordsjintweeti)or(thepropagationtimeisintervalbetweenthereplytwe
etiandthesourcetweet)thenXij=1otherwise0. Unlike other models, the authors constructed two
independent GCNs and then concatenated them into one fully connected layer for fake news
detection as Softmax((T⊕Re)W+b), where ⊕ is the concatenation function.

Vu et al. [125] presented a novel method called GraphSAGE for rumor detection based on
propagation detection. In contrast to other propagation-based approaches, this method
proposes a graph propagation embedding method based on a GCN to convert the news
propagation procedure and their features into vector space by aggregating the node feature
vectors and feature vectors of their local neighbors into a combination vector. Thus, the
difference between the GraphSAGE model and the traditional GCN models concerns the
aggregator functions, which are divided into the following aggregators: (i) Convolutional
aggregator:
 hkvj=σ(Wkhk−1vj+∑ihk−1vi|N(vj)|+1),∀vi∈N(vj)
(16)

(ii) LSTM aggregator:

△lstmk=LSTM({hk−1vi,∀vi∈N(vj)})
(17)

(iii) Pooling aggregator:

△poolk=max(σpool(Wpoolhk−1vi+bpool)),∀vi∈N(vj)
(18)

(iv) Linear aggregator:

△lineark=∑iwihk−1vi∑iwi,∀vi∈N(vj)
(19)
where wi is the weight vector of neighbor hvi. GraphSAGE efficiently integrates features such as
content, social, temporal, and propagation structures. These features can aggregate significant
information to train an algorithm for determining whether news are rumors or not. However,
this method requires data on the entire propagation procedure of the posted news. In some
cases, if the posted news does not obtain response opinions when it is spread, the accuracy of
the GraphSAGE model can be reduced.
Bian et al. [16] proposed a Bi-GCN model with two propagation operations, top-down (TD-GCN)
and bottom-up (BU-GCN), to detect two essential characteristics of rumors, dispersion, and
propagation. Bi-GCN was constructed as follows: (i) High-level node representations
as H(k)TD=σ(AˆTDH(k−1)W(k)TD) and H(k)BU=σ(AˆBUH(k−1)W(k)BU). (ii) Root feature
enhancement as follows H˜(k)TD=concat(H(k)TD,(H(k−1)TD)root) and H˜(k)BU= concat(H(k)BU,
(H(k−1)BU)root) where concat is a concatenate function; root indicates the root node; (iii) Node
representations are fed into the pooling aggregator
as STD=mean(H˜(2)TD) and SBU=mean(H˜(2)BU), and then concatenated them into one fully
connected layer for fake news detection as yˆ=Softmax(concat(STD,SBU)). This model can
capture both the propagation of rumor patterns using the TD-GCN and the dispersion of rumor
structures using the BU-GCN. Additionally, hidden information of the news is extracted through
layers of GCN to increase the influence of rumor roots. However, TD-GCN and BU-GCN were still
constructed independently.
Bai et al. [154] constructed a graph called an SR graph, where the node feature matrix X is
determined by word vectors; and the adjacent matrix A is defined as
follows: iftweeti repliestotweetj, thenAij=1otherwise0. Using the SR-graph, the authors
proposed an EGCN model for rumor detection as PG=H(k)=σ(AˆH(k−1)W(k)) with a node
proportion allocation mechanism as PT=TextCNN(A,X) where TextCNN indicates a conventional
CNN model. Let n and m be the number of nodes in the current SR-graph and the max SR-
graph, respectively, we have the feature output of the EGCN by Y=PG×nm+PT(1−nm). And the
output of the EGCN is determined by yˆ=Softmax(FC(Y)). This model focuses on exploiting the
impact of news content on the propagation procedure of rumors. However, the EGCN requires
data on the entire conversation regarding the posted news. In some cases, if the posted news
does not obtain response opinions when it is spread, its accuracy can be reduced.
Multidepth GCNs introduced by Hu et al. [190] combine the similarity of news to distinguish
them as fake or real via degrees of differences. This method can solve the significant challenge
of fake news detection, which is automatic fake news detection for short news items with
limited information, for example, headlines. Instead of stacking the GCN layer to merge
information over a long distance, the authors computed the different distance proximity
matrices to describe the relationship between nodes and explicitly protect the multigranularity
information, thus improving the node representation process with the diversity information.
Therefore, it performed k-step proximity to create different depth proximity matrices before
feeding to the GCN. For the step k-th proximity, the output is defined
as zk=AˆkReLU(AˆkXW(0)k)W(1)k,k=1,2,3,…, where node feature matrix X contains word
embeddings and representation of credit history. Aˆk is the k-th proximity matrix
as Aˆk=Aˆ×Aˆ×⋯×Aˆk,
where Aij=1ifi,jhave thesamejob−title,otherwiseAij=0. Then, the multi-depth information is
aggregated to create final representation Pj=∑mi=1αizi using an attention mechanism
as αi=exp(ui)∑ml=1exp(ul), where ui=tanh(Wizi+bi).
Nguyen et al. [155] introduced two methods, textual-based and graph-based, for false news
detection regarding COVID-19 and 5G conspiracy. For the first method, the authors detected
fake news using a combination of a pretrained BERT model and a multilayer perceptron model
to capture both the textual features and metadata of tweets. For the second method, the
author used a GCN with nine extracted features at each node, such as page rank, hub, and
authority, for content-based fake news detection. After implementing the two methods, the
authors proved that the performance of the first approach is better than that of the second.
Thus, metadata play a significant role in fake news detection. Therefore, improving the
efficiency of fake news detection by extracting metadata features for GCN should be considered
in the future. Regarding COVID-19 and 5G conspiracies, Pehlivan [194] introduced structure-
based fake news detection to evaluate the performance of existing models. Unlike other
methods, the author used only the temporal features of networks without considering textual
features. Two state-of-the-art models were selected to evaluate the GCN and DGCN [205].
Additionally, the authors used their temporal features to test the multivariate long short-term
memory fully convolutional network method [206]. Node feature matrices of the GCN and
DGCN are created based on the following values: degree centrality, closeness centrality,
betweenness centrality, load centrality, harmonic centrality, #cliques, clustering coefficient,
square clustering coefficient, and average neighbor degree. The node feature matrix of the
multivariate long short-term memory fully convolutional network is created with the average
clustering coefficient, #graph cliques, #connected components, local efficiency, #isolates, and
normalized time distance to the source tweet. Li et al. [191] introduced a propagation-based
method for determining political perspective by focusing on the social contextual information
of news. In this study, the GCN is used with an adjacency matrix to capture and represent the
social context via feature extractions: sharing actions, following actions regarding political
news, and a node feature matrix are used to capture the hidden content of news via word
embeddings. Meyers et al. [153] showed the significant role of propagation features in fake
news detection models. The authors first constructed a propagation graph to present important
information and then used a random forest classifier to train the graph and create node
embeddings. Finally, the GCN model was used to predict the authenticity of tweets. Unlike
other propagation graphs, the authors constructed the following graph: Let G={V,E} denote the
propagation graph, where V is a set of nodes including tweet nodes and retweet nodes and E is
a set of edges connected between a tweet node and its retweet node with a time weight. Thus,
this propagation graph includes a set of subgraphs, where each subgraph includes a tweet node
and its retweet nodes, and its depth never exceeds 1.

A social spammer detection model [201] was built with a combination of the GCN and Markov
random field (MRF) models. First, the authors used convolutions on directed graphs to explicitly
consider various neighbors. They then presented three influences of neighbors on a user’s label
(follow, follower, reciprocal) using a pairwise MRF. Significantly, the MRF is formulated as an
RNN for multistep inference. Finally, MRF layers were stacked on top of the GCN layers and
trained via an end-to-end process of the entire model. Unlike conventional GCNs, this model
uses an improved forward propagation rule

Q=H(l+1)=σ(D−1iAiH(l)W(l)i+D−1oAoH(l)W(l)o+Dˆ−1/2bAˆbDˆ−1/2bH(l)W(l)b)
(20)
where Ai,Ao,Ab are types of neighbors; Aˆb=Ab+I; Di,Do, and Dˆb are degree matrices of Ai,Ao,
and Ab, respectively; The node feature matrix X=H(0) is created based on BoW features. Then,
the authors initialized the posterior probabilities of the MRF layer with the GCN output as
R=Softmax(logH(k)−AiQ[−w−ww′−w]−AoQ[−ww′−w−w]−AbQ[−ww′w′−w])
(21)
where w,w′≥0 are two learnable parameters to measure homophily and heterophily strength of
MRF model. This method demonstrated the superiority of the combination of GCN and MRF
layers. A multistep MRF layer is essential to convergence. However, the node feature matrix
was created simply with the bag-of-words method. This limitation can be improved using state-
of-the-art embedding models in the future.
A novel GCN framework, called FauxWard [149], is proposed for fauxtography detection by
exploiting news characteristics, such as linguistic, semantic, and structural attributes. The
authors modeled fauxtography detection as a classification problem and used GCNs to solve
this problem. FauxWard is similar to traditional GCN models; however, unlike these models, it
adds a cluster-based pooling layer between graph convolutional layers to learn the node
representation more efficiently. The cluster-based pooling layer first assigns neighbor nodes
into clusters based on the node vectors of the previous graph convolution layers and then
learns a cluster representation as the input of the back graph convolution layer. It performs
graph convolution by A˜(k)=C(k−1)⊤A˜(k−1)C(k−1), where A˜(k) is the updated adjacency
matrix; C(k) is the clustering matrix obtained after the k-th graph convolution layer, such
that H(k)=C(k−1)⊤σ(A˜(k−1)H(k−1)W(k−1)), where H(0)=X be a node feature matrix. Unlike
conventional GCNs, this X is created by concatenating text content, such as linguistic,
sentiment, endorsement, and image content, such as metadata.
Malhotra et al. [147] introduced a method of combining RoBERTa and BiLSTM
(TD=Bi(RoTa(tweet)), where RoTa indicates a RoBERTa model [207] and Bi indicates a BiLSTM
model) and GCN methods (GD=H(k)= σ(AˆH(k−1)W(k)), where H(0)=X is a node feature matrix
by concatenating eleven features, such as friend count, follower count, followee count, etc.) for
rumor detection as yˆ=Softmax (concat(TD,GD)). This model is based on rumor characteristics,
such as propagation and content. It exploits features regarding the structure, linguistics, and
graphics of tweet news.
Vlad et al. [195] produced a multimodal multitask learning method based on two main
components: meme identification and hate speech detection. The first combines GCN and an
Italian BERT for text representation, whereas the second is an image representation method,
which varies among different image-based structures. The image component employed VGG-16
with five CNN stacks [208] to represent images. The text component used two mechanisms to
represent text, namely, Italian BERT attention and convolution. This model is multimodal
because it considers features related to the text and image content simultaneously. Meanwhile,
Monti et al. [3] introduced a geometric deep learning-based fake news detection method by
constructing heterogeneous graph data to integrate information related to the news, such as
user profile and interaction, network structure, propagation patterns, and content. Given a
URL u with a set of tweets mentioned u, the authors constructed a graph Gu={V,E}. V is a set of
nodes corresponding to tweets and their posters. E is a set of edges expressing one of four
relations between two nodes: follow, followed, spreading, and spread. This graph has node
feature matrix X and adjacency matrix A. X is created by characterizing user features, such as
profiles, network structure, and tweet content. However, matrix A is defined as
follows:if(nodevjspreadstweetofnodevi)or(nodevispreadstweetofnodevj)or(nodevifollowsnodev
j)or(nodevjfollowsnodevi)thenAij=1otherwise0; Given matrices X and A, similar to traditional
GCNs, the authors utilized a four-layer GCN: two convolutional layers for node representation
and two fully connected layers to predict the news as fake or real. However, unlike some
previous GCNs, in this proposal, one attention mechanism in the filters [178] and the mean
pooling are used to decrease the feature vectors’ dimension for each convolutional layer.
SELU [209] is employed as a nonlinearity activation function for the entire network.
Li et al. [123] presented a GCN-based antispam method for large-scale advertisements named
GAS. Unlike previous GCNs, in the GAS model, a combination graph is constructed by
integrating the nodes and edges of the heterogeneous graph and a homogeneous graph to
capture the local and global comment contexts. The GAS is defined in the following steps: (i)
Graphs construction: The authors constructed two types of graphs named Xianyu graph and
comment graph. The first graph was denoted by G = {U,I,E} where U,I are sets of nodes
representing users and their items, respectively, and E is a set of edges representing comments.
An adjacency of this graph is created as
follows: ifuserimakescommentetoitemj, thenAXij=1other− wise0. The second graph is
constructed by connecting nodes expressing comments that have the similar meaning. That
means ifcommentihassimilarmeaning withj,then ACij=1otherwise0. (ii) GCN on Xianyu graph:
Let h(l)e,h(l)U(e) and h(l)I(e) be the l-th layer node embeddings of edge, user, and item,
respectively, ze=h(l)e=σ(W(l)E⋅concat(h(l−1)e,h(l−1)U(e),h(l−1)I(e))) where h(0)e=TN(w0,w1,
…,wn) and U(e),I(e) are user node and item node of edge e. Let h(l)N(u),h(l)N(i) are neighbor
embeddings of node u,i. TN stands by TextCNN model [210]. wk is the word vector of word k in
tweet. Hence,
h(l)N(u)=σ(W(l)U⋅att(h(l−1)u,concat(h(l−1)e,h(l−1)i)))
(22)
where ∀e=(u,i)∈E(u) and
h(l)N(i)=σ(W(l)I⋅att(h(l−1)i,concat(h(l−1)i,h(l−1)e))),
(23)
where ∀e=(u,i)∈E(i), and E(u) is the edge connected to u; att a stands of attention mechanism.
From that, we
have: zu=h(l)u=concat(W(l)U⋅h(l)u,h(l)N(u)) and zi=h(l)i=concat (W(l)I⋅h(l)i,h(l)N(i)). (iii) GCN on
the comment graph: in this step, authors used the GCN model proposed in [211] to represent
nodes on the comment graph into node embeddings as pe=GCN(XC,AC), where XC is node
feature matrix. (iv) GAS classifier: The output of GAS model is defined
as y=classifier(concat(zi,zu,ze,pe)).

5.3. Detection approach based on AGNNs

Ren et al. [17] introduced a novel approach, called AA-HGNN, to model user and community
relations as a heterogeneous information network (HIN) for content-based and context-based
fake news detection. The primary technique used in AA-HGNN involves improving the node
representation process by learning the heterogeneous information network. In this study, the
AGNNs use two levels of an attention mechanism: the node learns the same neighbors’ weights
and then represents them by aggregating the neighbors’ weights corresponding to each type-
specific neighbor and a schema to learn the nodes’ information, thus obtaining the optimal
weight of the type-specific neighbor representations. Assume that we have a news HIN and a
news HIN schema, denoted by G={V,E} and SG={VT,ET}.
Let V={C∪N∪S} with C (creators), N (news), S (subjects); and E={Ec,n∪En,s}.
Let VT={θn,θc,θs} and ET={write,belongsto} denotes types of nodes and types of links. Node-
level attention is defined as h′ni=Mθn⋅hni, where ni∈N, hni is the feature vector of
node ni. Mθn is the transformation matrix for type θn. Let T∈{C∪N∪S}, tj∈T belongs to type-
neighbor θt and tj∈neighborni. Let eθtij=att(h′ni,h′tj;θt) is the importance degree of
node tj for ni, where att be a node-level attention mechanism with the attention weight
coefficient as αθtij=Softmaxj(eθtij) Then, the schema node is calculated by aggregating from the
neighbor’s features as Tni=σ(∑tj∈neighborniαθtij⋅h′tj). Let ωθti=schema(WTni,WNni) is the
importance degree of schema node Tni, where schema be a schema-level attention
mechanism, Nni is a schema node corresponding to neighbors of node ni. And the final fusion
coefficient is calculated as βθti=Softmaxt(ωθti). From that, we have a node representation
as rni=∑θt∈VTβθti⋅Tni. AA-HGNN can still achieve excellent performance without using much-
labeled data because it benefits from adversarial active learning. It can also be used for other
actual tasks relating to heterogeneous graphs because of its high generalizability.
Benamira et al. [192] proposed content-based fake news detection methods for binary text
classification tasks. The objective was a GNN-based semisupervised method to solve the
problem of labeled data limitations. This method comprises the following steps: news
embedding; news representation based on k nearest-neighbor graph inference; and news
classification based on GNNs, such as AGNN [179] and GCN [164], which are conventional GNNs
without improvements or updates.

5.4. Detection approach based on GAEs

Using the autoencoder special graph data, Kipf [170] used GAE to encode graphs to represent
latent structure information in graphs. GAEs are used in various fields, such as recommendation
systems [212] and link prediction [213], with reasonable performance. Recently, researchers
have begun to apply GAEs for fake news detection. The previous studies for fake news
detection models based on GAEs are summarized in Table 10.
Table 10

Comparison of AGNN and GAE methods.

Method
Critical idea Loss function Advantage Disadvantage
[Ref]
– Focus on analyzing
– Have not been
the news content – Can obtain good
evaluated with
Benamira using semi-supervised Cross-entropy efficacy with
big data and
et al. [192] learning loss limited labeled
multi-labeled
– Binary classification data
data
model
– The first model using
adversarial active – Support early
learning for fake news detection stage
detection – Still obtain high
– Improve the performance with – Not compare
AA-HGNN conventional GCN Cross-entropy limited training the efficacy with
[17] models by using a new loss data the context-
hierarchical attention – Can extract based methods
mechanism for node information as text
representation and structure
– Multi classification simultaneously
model
Lin – Focus on integrating The sum of – The first GAE- – Low
et al. [124] the information Cross-entropy based rumor performance for
related to text, loss and KL detection method the non-rumor
Method
Critical idea Loss function Advantage Disadvantage
[Ref]
propagation, and – Can create better
network structure and high-level
class
– Include three parts: node
divergence – Not high
encoder, decoder, and representations
loss generalization for
detector – Obtain better
the performance
– Multi classification efficacy than other
model the latest methods
Open in a separate window

Lin et al. [124] proposed a model to capture textual, propagation, and structural information
from news for rumor detection. The model includes three parts: an encoder, a decoder, and a
detector. The encoder uses a GCN to represent news text to learn information, such as text
content and propagation. The decoder uses the representations of the encoder to learn the
overall news structure . The detector also uses the representations of the encoder to predict
whether events are rumors. The decoder and detector are simultaneously implemented. These
parts are generally defined as follows: (i) Encoder component: Two layers of the GCN are used
to enhance the learning ability:

H(1)=GCN(X,A)=Aˆσ(AˆXW(0)W(1))
(24)

and
H(2)=GCN(H(1),A)=Aˆσ(AˆH(1)W(1)W(2))
(25)
where σ is ReLU function. X represents word vectors that are created by determining the TF-IDF
values, and the adjacent matrix A is defined as
follows: ifnodevirespondsto nodevj,thenAij=1otherwise0. Then, the GCN is used to learn a
Gaussian distribution for variational GAE as z=μ+ϵσ,
where μ=GCN(H(1),A) and logσ=GCN(H(1),A) (μ, σ, and ϵ are the mean, standard deviation, and
standard sample of the Gaussian distribution, respectively). (ii) Decoder component: In this
step, an inner product (⊙) is used to reconstruct the adjacency matrix as A˜=⊙(ZZ⊤) where Z is
the matrix of distributions z. (iii) Detector component: This step aims to represent the latent
information and classify the news. It is defined as S=MP(Z) where MP stands for the mean-
pooling operator. Finally, the output layer of this model is defined
as yˆ=Softmax(SW+b) where W is the parameter matrix of the fully connected layer.

6. Discussion
6.1. Discussion on GNNs∗-based methods
Previous studies for fake news detection models based on GNNs∗ are compared in Table 7.
Table 7

Comparison of GNNs∗ methods.

Method [Ref] Critical idea Loss function Advantage Disadvantage

– Focus on
analyzing the
news content – Have not been
– Can obtain high
using semi- evaluated with
Benamira Cross-entropy efficacy with
supervised big data and
et al. [192] loss limited labeled
learning multi-labeled
data
– Binary data
classification
model
– Propagation-
based fake news – Can improve
detection method the performance
– Ignore the
– Focus on of conventional
selection of
continual learning Elastic Weight methods without
features or the
Yi Han et al. [4] and incremental Consolidation using any text
finding of
training loss information
“universal”
techniques – Can handle
features
– Use two unseen data and
techniques: EWC new data
and GEM
– Can use to
multi-
classification
– Focus on two – Tasks
tasks
tasks aiming to implement
– Binary
analysis and separately
classification task
FakeNews [193] detect fake news NA corresponding
obtains
via news textual to information
significantly
and news textual and
higher
structure structure
performance
than the ternary
ones
SAFER [15] – Contextual fake NA – Improve the – Sensitive to
news detection performance of bottleneck and
method the traditional over-smoothing
– Focus on GNNs problems
 Method [Ref] Critical idea Loss function Advantage Disadvantage
combining
information: – Can add more
content nature, layers to identify
user behaviors, more efficacy
and users social neighborhood
network
– Improve – Stance
– Contextual fake
representation detection and
news detection The total of
quality textual encode
method unsupervised
– Can use to a have not been
– Focus on proximity loss,
limited training jointly optimized
FANG [5] representation self-super-vised
dataset – Sharing
quality by stance loss, and
– Can capture contents and
capturing sharing supervised fake
temporal hyperlinks
patterns and news loss
patterns of fake become
social structure
news obsolete quickly
Open in a separate window
We presented the main steps, advantages, and disadvantages of GNN∗-based methods for fake
news detection. Some of our assessments are as follows: Regarding the extracted
features, [4] used only user-based features; [5] used features based on networks, users, and
linguistics; and [193] used linguistic-based features (textual analysis). Meanwhile, [15] used
features related to networks and linguistics. Regarding graph
structure, [4], [5], [193] constructed a homogeneous graph. However, unlike [4], [193] only one
graph was constructed, and [5] created two subgraphs to represent news sources and news
users. Meanwhile, [15] built a heterogeneous graph with two types of nodes and edges.
However, although the graph structure of [15] is better than that of the other three
models, [192] provides the best performance. This result may be because [5] can better extract
meaningful features in fake news detection. Therefore, to develop new GNN∗-based models in
the future, more attention should be given to extracting excellent features and building good
standard data instead of focusing on improving the graph structure.

6.2. Discussion on GCNs-based methods

Previous studies for fake news detection models based on GCNs are compared in Table
8, Table 9.
Table 8

Comparison of GCN methods.

Loss
Method [Ref] Critical idea Advantage Disadvantage
function
– Focus on
analyzing the news – Enable to
content, users, integrate – Only
social structure heterogeneous implement with
Monti et al. [3] and propagation Hinge loss data the binary
using geometric – Obtain very high classification
deep learning performance with task
– Binary big real data
classification task
– Focus on
– Can solve spam
capturing the
problems like
global and local
adversarial
contexts of the – Only
actions and
news implement with
Regression scalability
!GAS [123] – Integrate graphs the binary
loss – Obtain high
of homogeneous classification
performance with
and task
large-scale data
heterogeneous
– Can apply to
– Binary
online news
classification task
– Focus on
capturing the – Use a very
propagation broad definition
features of the – Can apply to the – Apply to a
Marion et al. [153] news using NA non-URL-limited single data
geometric deep news source
learning – Not high
– Binary generalizability
classification task
GCAN [14] – Focus on the Cross- – Can early – Apply to a
relation of original entropy detection single data
tweet and retweet loss – Can detect a source
and the co- tweet story as – Not high
influence of the fake using only generalization
user interaction short-text tweet
and original tweet without needing
 Loss
Method [Ref] Critical idea Advantage Disadvantage
function
– Use the dual co- user comments
attention and network
mechanism structure
– Binary – Explainable of
classification task fake reasons
– Focus on the
– Not promising
temporal features
performance
of the network
Cross- – The data is
structure without – Can apply to
Pehlivan et al. [194] entropy split not
considering any metadata
loss reasonable for
textual features
training, testing,
– Binary
validation
classification task
– Focus on
analyzing features
– Have an early
related to
detection
dispersion and
mechanism
propagation of the
– Can detect
news
Cross- rumors in real-
– Construct a top- – Not high
*Bi-GCN [16] entropy time
down graph to generalization
loss – Obtain much
learn rumor spread
higher
and a bottom-up
performance than
graph to capture
state-of-the-art
rumor dispersion
methods
– Multi
classification task
*GCNSI [196] – Focus on Sigmiod – First model – Have to
identifying cross- based on multiple retrain the
multiple sources of entropy sources of the model if the
rumor without any loss rumor graph structure
knowledge related – Improve the is changed
to news performance of – Take quite
propagation the state-of-the- much time to
– Improve the art methods by train and obtain
previous GCN about 15% suitable
models by parameters
modifying the
enhanced node
representations
 Loss
Method [Ref] Critical idea Advantage Disadvantage
function
and loss function
– Multi
classification task
– The first semi-
supervised model
focus on
continuously
integrating both
methods of
feature-based and
– Obtain superior – Use simple
propagation-based Cross-
!GCNwithMRF [201 effectiveness BoW for
– Use the deep entropy
] – Can ensure features
learning model loss
convergence representation
with a refined MRF
layer on directed
graphs to enable
the end-to-end
training
– Multi
classification task
Open in a separate window

Table 9

Comparison of GCN methods (continued).

Loss
Method [Ref] Critical idea Advantage Disadvantage
function
– Focus on
combining features
related to text and
users – Evaluated by a
Cross- – Enable for more
*Malhotra – Use the geometric limited dataset
entropy efficiently features
et al. [147] deep learning with – Overfitting test
loss extraction
RoBERTa-based error
embedding
– Multi classification
task
!FauxWard [149] – Focus on features Cross- – Obtain a – Not directly
related to the entropy significant analyze the
 Loss
Method [Ref] Critical idea Advantage Disadvantage
function
linguistic and
semantic of the user
comments and the
user network
structure performance content of the
– Use the geometric loss within a short news containing
deep learning on a time window an image-centric
user comment
network
– Binary
classification model
– Focus on depth
integrating of
contextual
– Have an early
information and
detection
propagation
mechanism
structure – Random split for
– Can create a
– Use multi-head Cross- validation data
better semantic-
KZWANG [156] attention entropy and manual split
integrated
mechanism to loss for training,
representation
create contextual testing data
– Improve
representation
performance
without extracting
significantly
any features
– Multi classification
model
*GraphSAGE [125] – Focus on Cross- – High – Can reduce the
determining entropy generalization for performance if
patterns loss unseen data not use full
propagation-based – Reduce the information of
characteristics and detection error of post (original and
information related state-of-the-art response) in the
to the content, methods down to spread process
social network 10%
structure, and delay – Efficiently
time integrate features
– Use a graph related to the
embedding whole propagated
technique to post
integrate
 Loss
Method [Ref] Critical idea Advantage Disadvantage
function
information of graph
structure and node
features
– Multi classification
model
– Focus on using
features related to – Can create – Not high
the content of news better word generalization
text representations – No suitable
Bert-GCN – Improve the other – Can improve the augmentation
NA
Bert-VGCN [150] GCN-based models performance of data to improve
using BERT-based the conventional features
embeddings GCN method extraction and
– Multi classification significantly avoid overfitting
model
– Focus on
information of text
content, spread – Obtain high
– Strong depend
time, social network efficacy in early
on the full
structure detection
Cross- information of
– Construct – Can improve the
*Lotfi [204] entropy both original
weighted graphs performance of
loss tweet and
based on users the state-of-the-
response tweets
interaction in art methods
of conversations
conversations significantly
– Binary
classification model
– Focus on capturing
the information of – Optimal capture
users interactions of user’s – Not high-
– Improve the interactions performance
Cross-
conventional GCN – Capture better generalization due
*SAGNN [151] entropy
models by adding the different to only comparing
loss
one or more features between with one baseline
aggregation layer rumors and non- method
– Multi classification rumors
model
*EGCN [154] – Focus on fully NA – Can obtain – Not high
extracting features comparable generalization
related to text performance or
 Loss
Method [Ref] Critical idea Advantage Disadvantage
function
content and
better than
structure
machine learning
– Construct
methods
weighted graphs of
– Can use the
source-replies
information of the
relation for
global and local
conversations
structure
– Binary
simultaneously
classification model
Open in a separate window

We presented the main steps, advantages, and disadvantages of GCN-based methods for fake
news detection. In our assessments, methods such as [3], [14], [16], [123], [191], and [196],
show the best efficiency, where two methods are used for fake news detection, two for rumor
detection, and two for spam classification. Regarding the two papers in the first
category, [3] was the first to apply GCNs for fake news detection. This method focuses on
extracting user-based, network-based, and linguistic-based features to build propagation-based
heterogeneous GCNs. The authors determined that this proposal can obtain a more promising
result than content-based methods. Conversely, [14] is an enriched GCN with a dual coattention
mechanism. This method uses user-based and linguistic-based features to construct
homogeneous GCNs with a dual coattention mechanism. In our assessment, although [14] used
dual coattention mechanisms, the efficiency was still lower than that in [3]. Noticeably, this
result is attributable mainly to more features being extracted by [3] than by [14]. Additionally,
the graph structure used in [3] was evaluated as better than the structure used in [14]. Moving
forward, we hope to improve the performance of fake news detection methods by building
dual coattention heterogeneous GCNs using user-based, network-based, and linguistic-based
features simultaneously. For the two papers in the second category, both methods were built
to detect rumors by propagation-based GCNs. The difference is that [16] constructed
bidirectional GCNs to capture the rumor dispersion structure and rumor propagation patterns
simultaneously. Meanwhile, [196] created unidirectional GCNs based on the information of
multiorder neighbors to capture rumor sources. In our view, [16] can outperform [196] because
rumor detection, rumor propagation, and dispersion are more critical than rumor sources. For
the two papers in the last category, [123], [191] also proposed similar methods for spam
detection using social context-based GCNs. The different points are that [123] built a model
integrating heterogeneous and homogeneous graphs to capture both local and global news
contexts. In contrast, [191] constructed only one heterogeneous graph to capture the general
news context. In our opinion, the model presented in [123] is more comprehensible, can be
reimplemented, and yields slightly better results than the method in [191]. The reason for this
result is that building each type of graph is suitable for the capture and integration of each type
of context, which can capture the news context more comprehensively than constructing one
graph for all contexts. Thus, when building fake news detection models based on GNNs,
different graphs should be constructed to capture each specific type of information and then
perform the fusion step. This approach promises to provide better performance than building
one type of graph to capture all types of information. We maximally limit the construction of a
general graph and then divide it into specific types because the breakdown of the graph can
easily result in the loss of information on the relationship among edges.

6.3. Discussion on AGNNs- and GAEs-based methods

Previous studies for fake news detection models based on AGNNs and GAEs are compared
in Table 10.

We presented the main steps, advantages, and disadvantages of the two methods in the AGNN
and AGE categories for fake news detection. Evidently, [17] presented a more detailed fake
news detection method than [192]. Additionally, the method in [17] was proposed after that
in [192]; thus, it is better than [192]. For example, [192] constructed a homogeneous graph,
whereas [17] created a heterogeneous graph. The heterogeneous graph was evaluated as
superior to the homogeneous graph because it can capture more meaningful information.
Therefore, it obtains better results than [192]. Meanwhile, the Lin et al. [124] method uses a
conventional GCN variant to encode the latent representation of graphs. This method can
capture the entire structural information efficiency. It can thus enrich traditional GCNs by
adding two more components, namely, the decoder and detector. However, this study focused
only on user-based and linguistic-based features, ignoring network-based features; therefore,
the desired effect is not expected.

7. Challenges
7.1. Fake news detection challenges
Based on recent publications in the field of fake news detection, we summarized and classified
challenges into five categories, where each category of challenge corresponds to one category
of fake news detection. The details of each type of challenge are shown in Fig. 6. The following
presents significant challenges that can become future directions in fake news detection.
Fig. 6
List of challenges of fake news detection.

Deepfake [214] is a hyperrealistic, digitally controlled video that shows people saying or doing
things that never truly happened or composite documents generated based on artificial
intelligence techniques. Given the sophistication of these counterfeiting techniques,
determining the veracity of the public appearances or influencer claims is challenging owing to
fabricated descriptions. Therefore, Deepfake currently poses a significant challenge to fake
news detection.

The hacking of influencers’ accounts to spread fake news or disinformation about a speech by
celebrities themselves is also a unique phenomenon in fake news detection. However, this
information will be quickly removed when the actual owner of these accounts discovers and
corrects them. However, at the time of its spread, this information causes extremely harmful
effects. Instantly detecting whether the posts of influencers are fake has thus become an
important challenge.

News may be fake at one point in time and real at another. That is, the news is real or fake,
depending on the time it is said and spread. Therefore, real-time fake news detection has not
yet been thoroughly addressed.

Constructing benchmark datasets and determining the standard feature sets corresponding to
each approach for fake news detection remain challenges.
Kai Shu et al. [215] constructed the first fake news detection methods by effectively extracting
content, context, and propagation features simultaneously through four embedding
components: news content, news users, user-news interactions, and publisher news relations.
Then, these four embeddings were fed into a semisupervised classification method to learn a
classification function for unlabeled news. In addition, this method can be used for fake news
early detection. Ruchansky et al. [28] constructed a more accurate fake news prediction model
by extracting the behavior of users, news, and the group behavior of fake news propagators.
Then, three features were fed into the architecture, including three modules as follows: (i) use
a recurrent neural network to capture the temporal activity of a user on given news via news
and propagator behaviors; (ii) learn the news source via user behavior; and (iii) integrate the
previous two modules for fake news detection with high accuracy. From this survey of
literature, we see that the most effective approaches combine features regarding content,
context, and propagation. Although these combination methods may have high complexity
regarding the algorithms used, the many extracted features, and high feature dimensions, they
can simultaneously capture various aspects of fake news. Therefore, the most efficacious and
least costly extraction of content, propagation patterns, and users’ stance simultaneously is not
only a promising solution but also a significant challenge for fake news detection.
7.2. Challenges related to graph neural networks
Based on studying the related literature, this section summarizes some challenges of GNN-
based methods and then identifies possible future directions.
Most conventional GNNs utilize undirected graphs and edge weights as binary values (1 and
0) [216] unsuitable for many actual tasks. For example, in graph clustering, a graph partition is
sought that satisfies two conditions:

(i) the difference between the weights of edges among unlike groups is as low as possible; (ii)
the difference in the weights of edges among similar groups is as high as possible.
Here, if the weight of the edges is a binary value, the given problem cannot be solved using this
graph. Therefore, future studies can construct graphs with the weights of edges as the actual
values representing the relationship among the nodes as much as possible.

For NLP tasks, GNNs have not represented node features by capturing the context of a
paragraph or an entire sentence. Alternatively, these methods have also overlooked the
semantic relationships among phrases in the sentences. For example, for sentiment
classification tasks, we have the sentence “The smell of this milk tea is not very fragrant.” This
sentence includes a fuzzy sentiment phrase, namely “not very fragrant”. Some approaches
classify this sentence as expressing a positive sentiment because they only focus on “fragrant”,
ignoring the role of both “not” and “very”, whereas other models determine the expression as a
negative sentiment because they ignore the impact of “very”. Therefore, future directions for
improving GNN-based models should focus on determining node features based on sentence
embeddings or significant phrase embeddings.

Capturing context, content, semantic relations, and sentiment knowledge simultaneously in

sentences is essential for GNN-based NLP tasks. Meanwhile, only a few studies have
incorporated some of these features by flexible GNNs to improve the efficiency of NLP tasks,
including fake news detection. For instance, in [217], the authors extracted common sense
knowledge and syntax via GNNs, whereas in [218], the authors constructed a single text-based
GNN by representing document-word relations and word co-occurrence. To the best of our
knowledge, no GNN has simultaneously considered all the content, contexts, common sense
knowledge, and semantic relations. This task remains an exciting challenge for NLP task-based
GNNs.
So far, GCNs have been limited to a few layers (two or three) owing to the vanishing gradient,
which limits their real-world applications. For example, GCNs in [217], [219], [220] stopped at
two layers because of the vanishing gradient error. Therefore, constructing deep fuzzy GCNs of
syntactic, knowledge, and context by using the deep learning algorithm over the combination
graph of the fuzzy syntactic graph, the fuzzy knowledge graph and the fuzzy context graph can
solve the aforementioned limitations of previous methods for aspect-level sentiment analysis.
8. Conclusion and open issues

GNN-based fake news detection is relatively new. Thus, the number of published studies is
limited. Although we did not implement methods presented in the 27 studies on the same
datasets and did not evaluate their efficiency on the same comparison criteria, the 27 papers
surveyed here show that this method initially obtained excellent results. Additionally, many
challenges need to be addressed to achieve more comprehensive results, which we discussed
at the end of the corresponding sections. Nonetheless, given the 27 surveyed papers, promising
results are expected in the future. By addressing these challenges, we hope to improve the
effectiveness of GNN-based fake news detection. The following paragraphs analyze some
challenges for GNN-based fake news detection and discuss future directions.

Benchmark data: Recently, some researchers have argued that when training a system, data
affect system performance more than algorithms do [221]. However, in our assessment, we had
no graph benchmark data for fake news detection in the graph learning community. Graph-
based fake news detection benchmarks may present an opportunity and direction for future
research.

Compatible hardware: With the rapid growth of Deepfake, graphs to represent these data will
become more complex. However, the more scalable GNNs are, the higher the price and
complexity of the algorithms is. Scientists often use graph clustering or graph sampling to solve
this problem, ignoring the information loss of the graph using these techniques. Therefore, in
the future, graph scalability may be solved by developing dedicated hardware that fits the
graph structure. For example, GPUs were a considerable leap forward in lowering the price and
increasing the speed of deep learning algorithms.

Fake news early detection: Early detection of fake news involves detecting fake news at an
early stage before it is widely disseminated so that people can intervene early, prevent it early,
and limit its harm. Early detection of fake news must be accomplished as soon as possible
because the more widespread fake news is, the more likely it is that the authentication effect
will take hold, meaning that people will be likely to believe the information. Currently, for fake
news early detection, people often focus on analyzing the news content and the news context,
which leads to three challenges. First, new news often appears to bring new knowledge, which
has not been stored in the existing trust knowledge graph and cannot be updated immediately
at the time the news appears. Second, fake news tends to be written with the same content but
with different deceptive writing styles and to appear simultaneously in many various fields.
Finally, limited information related to news content, news context, news propagation, and
latent information can adversely affect the performance of GNN-based detection methods.

Dynamic GNNs: Most graphs used in the current GNN-based fake news detection methods
have a static structure that is difficult to update in real time. In contrast, news authenticity can
change continuously over time. Therefore, it is necessary to construct dynamic graphs that are
spatiotemporally capable of changing with real-time information.

Heterogeneous GNNs: The majority of current GNN-based fake news detection models
construct homogeneous graphs. However, it is difficult to represent all the news texts, images,
and videos simultaneously on these graphs. The use of heterogeneous graphs that contain
different types of edges and nodes is thus a future research direction. New GNNs are suitable
for heterogeneous graphs, which are required in the fake news detection field.

Multiplex GNNs: As analyzed in Section 7.2, most GNN-based fake news detection approaches
have focused on independently using propagation, content, or context features for
classification. Very few methods have used a combination of two of the three features. No
approach uses a hybrid of propagation, content, and context simultaneously in one model.
Therefore, this issue is also a current challenge in fake news detection. In the future, research
should build GNN models by constructing multiplex graphs to represent news propagation,
content, and context in the same structure.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal
relationships that could have appeared to influence the work reported in this paper.

Footnotes
1
[Link]
2
[Link]
3
[Link]
4
[Link]
5
[Link]
6
[Link]
7
[Link] research/lt/resources/satire/.
8
[Link] sfu-discourse-lab/MisInfoText.
9
[Link]
10
[Link]
11
[Link]

Appendix. Description of datasets

Fact-checking: The English dataset with 221 statements regarding society and politics was
collected from online streaming.

EMERGENT: The English dataset with 300 claims and 2595 associated article headlines
regarding society and technology were collected from online streaming and Twitter.

Benjamin Political News: The English dataset with 225 stories regarding politics was collected
from online streaming from 2014 to 2015.

Burfoot Satire News7 : The English dataset with 4233 news articles regarding economy, politics,
society, and technology was collected from online streaming.

MisInfoText8 : The English dataset with 1692 news articles regarding society was collected from
online streaming.

Ott et al.’s dataset: The English dataset with 800 reviews regarding tourism was collected from
TripAdvisor social media.

FNC-1: The English dataset with 49,972 articles regarding politics and society were collected
from online streaming.

Fake_or_real_news: The English dataset with 6337 articles regarding politics and society was
collected from online streaming.

TSHP-17: The English dataset with 33,063 articles regarding politics was collected from online
streaming.

QProp9 : The English dataset with 51,294 articles regarding politics was collected from online
streaming.

NELA-GT-201810 : The English dataset with 713,000 articles regarding politics was collected from
online streaming from February 2018 to November 2018.

TW_info: The English dataset with 3472 articles regarding politics was collected from Twitter
from January 2015 to April 2019.
FCV-2018: The dataset, including 8 languages with 380 videos and 77,258 tweets regarding
society, was collected from three social networks, namely YouTube, Facebook, and Twitter from
April 2017 to July 2017.

Verification Corpus: The dataset including 4 languages with 15,629 posts regarding 17 society
events (hoaxes) was collected from Twitter from 2012 to 2015.

CNN/Daily Mail: The English dataset with 287,000 articles regarding politics, society, crime,
sport, business, technology, and health was collected from Twitter from April 2007 to April
2015.

Tam et al.’s dataset: The English dataset with 1022 rumors and 4 million tweets regarding
politics, science, technology, crime, fauxtography, and fraud/scam was collected from Twitter
from May 2017 to November 2017.

FakeHealth11 : The English dataset with 500,000 tweets, 29,000 replies, 14,000 retweets, and
27,000 user profiles with timelines and friend lists regarding health were collected from Twitter.

Data availability

No data was used for the research described in the article

References
1. Bovet A., Makse H.A. Influence of fake news in Twitter during the 2016 US presidential
election. Nature Commun. 2019;10(1):1–14. [PMC free article] [PubMed] [Google Scholar]
2. Allcott H., Gentzkow M. Social media and fake news in the 2016 election. J. Econ.
Perspect. 2017;31(2):211–236. [Google Scholar]
3. Monti F., Frasca F., Eynard D., Mannion D., Bronstein M.M. 2019. Fake news detection on
social media using geometric deep learning. arXiv preprint arXiv:1902.06673. [Google Scholar]
4. Han Y., Karunasekera S., Leckie C. 2020. Graph neural networks with continual learning for
fake news detection from social media. arXiv preprint arXiv:2007.03316. [Google Scholar]
5. V.-H. Nguyen, K. Sugiyama, P. Nakov, M.-Y. Kan, Fang: Leveraging social context for fake news
detection using graph representation, in: Proceedings of the 29th ACM International
Conference on Information & Knowledge Management, 2020, pp. 1165–1174.
6. Wu Z., Pan S., Chen F., Long G., Zhang C., Philip S.Y. A comprehensive survey on graph neural
networks. IEEE Trans. Neural Netw. Learn. Syst. 2020;32(1):4–24. [PubMed] [Google Scholar]
7. J. Redmon, S. Divvala, R. Girshick, A. Farhadi, You only look once: Unified, real-time object
detection, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,
2016, pp. 779–788.
8. W. Shi, R. Rajkumar, Point-gnn: Graph neural network for 3d object detection in a point
cloud, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,
2020, pp. 1711–1719.
9. G. Chen, Y. Tian, Y. Song, Joint Aspect Extraction and Sentiment Analysis with Directional
Graph Convolutional Networks, in: Proceedings of the 28th International Conference on
Computational Linguistics, 2020, pp. 272–279.
10. Zhang C., Li Q., Song D. 2019. Aspect-based sentiment classification with aspect-specific
graph convolutional networks. arXiv preprint arXiv:1909.03477. [Google Scholar]
11. Bastings J., Titov I., Aziz W., Marcheggiani D., Sima’an K. 2017. Graph convolutional
encoders for syntax-aware neural machine translation. arXiv
preprint arXiv:1704.04675. [Google Scholar]
12. Marcheggiani D., Bastings J., Titov I. 2018. Exploiting semantics in neural machine
translation with graph convolutional networks. arXiv preprint arXiv:1804.08313. [Google
Scholar]
13. Zhang S., Tong H., Xu J., Maciejewski R. Graph convolutional networks: A comprehensive
review. Comput. Soc. Networks. 2019;6(1):1–23. [Google Scholar]
14. Lu Y.-J., Li C.-T. 2020. GCAN: Graph-aware co-attention networks for explainable fake news
detection on social media. arXiv preprint arXiv:2004.11648. [Google Scholar]
15. Chandra S., Mishra P., Yannakoudakis H., Nimishakavi M., Saeidi M., Shutova E. 2020.
Graph-based modeling of online communities for fake news detection. arXiv
preprint arXiv:2008.06274. [Google Scholar]
16. Bian T., Xiao X., Xu T., Zhao P., Huang W., Rong Y., Huang J. vol. 34. 2020. Rumor detection
on social media with bi-directional graph convolutional networks; pp. 549–556. (Proceedings of
the AAAI Conference on Artificial Intelligence). [Google Scholar]
17. Ren Y., Wang B., Zhang J., Chang Y. Adversarial active learning based heterogeneous graph
neural network for fake news detection. 2020 IEEE International Conference on Data Mining;
ICDM; IEEE; 2020. pp. 452–461. [Google Scholar]
18. Collins B., Hoang D.T., Nguyen N.T., Hwang D. Trends in combating fake news on social
media–A survey. J. Inf. Telecommun. 2021;5(2):247–266. [Google Scholar]
19. Khan T., Michalas A., Akhunzada A. Fake news outbreak 2021: Can we stop the viral
spread? J. Netw. Comput. Appl. 2021 [Google Scholar]
20. Klyuev V. Fake news filtering: Semantic approaches. 2018 7th International Conference on
Reliability, Infocom Technologies and Optimization (Trends and Future Directions); ICRITO; IEEE;
2018. pp. 9–15. [Google Scholar]
21. Oshikawa R., Qian J., Wang W.Y. 2018. A survey on natural language processing for fake
news detection. arXiv preprint arXiv:1811.00770. [Google Scholar]
22. Shu K., Bhattacharjee A., Alatawi F., Nazer T.H., Ding K., Karami M., Liu H. Combating
disinformation in a social media age. Wiley Interdiscipl. Rev.: Data Min. Knowl.
Discov. 2020;10(6) [Google Scholar]
23. Mahmud F.B., Rayhan M.M.S., Shuvo M.H., Sadia I., Morol M.K. A comparative analysis of
graph neural networks and commonly used machine learning algorithms on fake news
detection. 2022 7th International Conference on Data Science and Machine Learning
Applications; CDMA; IEEE; 2022. pp. 97–102. [Google Scholar]
24. Shu K., Sliva A., Wang S., Tang J., Liu H. Fake news detection on social media: A data mining
perspective. ACM SIGKDD Explor. Newsl. 2017;19(1):22–36. [Google Scholar]
25. Lazer D.M., Baum M.A., Benkler Y., Berinsky A.J., Greenhill K.M., Menczer F., Metzger M.J.,
Nyhan B., Pennycook G., Rothschild D., et al. The science of fake
news. Science. 2018;359(6380):1094–1096. [PubMed] [Google Scholar]
26. Quandt T., Frischlich L., Boberg S., Schatto-Eckrodt T. The International Encyclopedia of
Journalism Studies. John Wiley & Sons, Inc. Hoboken, NJ, USA; 2019. Fake news; pp. 1–
6. [Google Scholar]
27. Zhou X., Zafarani R. 2018. Fake news: A survey of research, detection methods, and
opportunities. arXiv preprint arXiv:1812.00315. [Google Scholar]
28. N. Ruchansky, S. Seo, Y. Liu, Csi: A hybrid deep model for fake news detection, in:
Proceedings of the 2017 ACM on Conference on Information and Knowledge Management,
2017, pp. 797–806.
29. K. Shu, L. Cui, S. Wang, D. Lee, H. Liu, defend: Explainable fake news detection, in:
Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data
Mining, 2019, pp. 395–405.
30. Cinelli M., Morales G.D.F., Galeazzi A., Quattrociocchi W., Starnini M. The echo chamber
effect on social media. Proc. Natl. Acad. Sci. 2021;118(9) [PMC free article] [PubMed] [Google
Scholar]
31. Paul C., Matthews M. The Russian “firehose of falsehood” propaganda model. Rand
Corporation. 2016;2(7):1–10. [Google Scholar]
32. Del Vicario M., Bessi A., Zollo F., Petroni F., Scala A., Caldarelli G., Stanley H.E.,
Quattrociocchi W. 2015. Echo chambers in the age of misinformation. arXiv
preprint arXiv:1509.00189. [Google Scholar]
33. Del Vicario M., Bessi A., Zollo F., Petroni F., Scala A., Caldarelli G., Stanley H.E.,
Quattrociocchi W. The spreading of misinformation online. Proc. Natl. Acad.
Sci. 2016;113(3):554–559. [PMC free article] [PubMed] [Google Scholar]
34. Del Vicario M., Vivaldo G., Bessi A., Zollo F., Scala A., Caldarelli G., Quattrociocchi W. Echo
chambers: Emotional contagion and group polarization on Facebook. Sci. Rep. 2016;6(1):1–
12. [PMC free article] [PubMed] [Google Scholar]
35. Egelhofer J.L., Lecheler S. Fake news as a two-dimensional phenomenon: A framework and
research agenda. Ann. Int. Commun. Assoc. 2019;43(2):97–116. [Google Scholar]
36. Bakir V., McStay A. Fake news and the economy of emotions: Problems, causes,
solutions. Digit. Journalism. 2018;6(2):154–175. [Google Scholar]
37. Franklin B., McNair B. Routledge; 2017. Fake News: Falsehood, Fabrication and Fantasy in
Journalism. [Google Scholar]
38. Tandoc Jr. E.C., Lim Z.W., Ling R. Defining “fake news” a typology of scholarly
definitions. Digit. Journalism. 2018;6(2):137–153. [Google Scholar]
39. Wardle C. Fake news. It’s complicated. First Draft. 2017;16:1–11. [Google Scholar]
40. Stahl K. Fake news detection in social media. Calif. State Univ. Stanislaus. 2018;6:4–
15. [Google Scholar]
41. Hanson N.R. A note on statements of fact. Analysis. 1952;13(1):24. [Google Scholar]
42. Pierri F., Ceri S. False news on social media: A data-driven survey. ACM Sigmod
Record. 2019;48(2):18–27. [Google Scholar]
43. Vosoughi S., Roy D., Aral S. The spread of true and false news
online. Science. 2018;359(6380):1146–1151. [PubMed] [Google Scholar]
44. Kshetri N., Voas J. The economics of “fake news” IT Prof. 2017;19(6):8–12. [Google Scholar]
45. Fox E.J., Hoch S.J. Cherry-picking. J. Mark. 2005;69(1):46–62. [Google Scholar]
46. Zubiaga A., Aker A., Bontcheva K., Liakata M., Procter R. Detection and resolution of
rumours in social media: A survey. ACM Comput. Surv. 2018;51(2):1–36. [Google Scholar]
47. Allen F., Gale D. Stock-price manipulation. Rev. Financ. Stud. 1992;5(3):503–529. [Google
Scholar]
48. Rubin V.L., Chen Y., Conroy N.K. Deception detection for news: Three types of fakes. Proc.
Assoc. Inf. Sci. Technol. 2015;52(1):1–4. [Google Scholar]
49. Y. Chen, N.J. Conroy, V.L. Rubin, Misleading online content: Recognizing clickbait as” false
news”, in: Proceedings of the 2015 ACM on Workshop on Multimodal Deception Detection,
2015, pp. 15–19.
50. Hofmann B. Fake facts and alternative truths in medical research. BMC Med.
Ethics. 2018;19(1):1–5. [PMC free article] [PubMed] [Google Scholar]
51. Gentzkow M., Shapiro J.M., Stone D.F. vol. 1. Elsevier; 2015. Media bias in the marketplace:
Theory; pp. 623–645. (Handbook of Media Economics). [Google Scholar]
52. D’Ulizia A., Caschera M.C., Ferri F., Grifoni P. Fake news detection: A survey of evaluation
datasets. PeerJ Comput. Sci. 2021;7 [PMC free article] [PubMed] [Google Scholar]
53. Sharma K., Qian F., Jiang H., Ruchansky N., Zhang M., Liu Y. Combating fake news: A survey
on identification and mitigation techniques. ACM Trans. Intell. Syst. Technol. 2019;10(3):1–
42. [Google Scholar]
54. Ahmed H., Traore I., Saad S. International Conference on Intelligent, Secure, and Dependable
Systems in Distributed and Cloud Environments. Springer; 2017. Detection of online fake news
using n-gram analysis and machine learning techniques; pp. 127–138. [Google Scholar]
55. Ahmed H., Traore I., Saad S. Detecting opinion spams and fake news using text
classification. Secur. Privacy. 2018;1(1) [Google Scholar]
56. Nakamura K., Levy S., Wang W.Y. 2019. R/fakeddit: A new multimodal benchmark dataset
for fine-grained fake news detection. arXiv preprint arXiv:1911.03854. [Google Scholar]
57. Wang W.Y. 2017. ” Liar, liar pants on fire”: A new benchmark dataset for fake news
detection. arXiv preprint arXiv:1705.00648. [Google Scholar]
58. Shu K., Mahudeswaran D., Wang S., Lee D., Liu H. Fakenewsnet: A data repository with news
content, social context, and spatiotemporal information for studying fake news on social
media. Big Data. 2020;8(3):171–188. [PubMed] [Google Scholar]
59. J. Golbeck, M. Mauriello, B. Auxier, K.H. Bhanushali, C. Bonk, M.A. Bouzaghrane, C. Buntain,
R. Chanduka, P. Cheakalos, J.B. Everett, et al., Fake news vs satire: A dataset and analysis, in:
Proceedings of the 10th ACM Conference on Web Science, 2018, pp. 17–21.
60. F.K.A. Salem, R. Al Feel, S. Elbassuoni, M. Jaber, M. Farah, Fa-kes: A fake news dataset
around the Syrian war, in: Proceedings of the International AAAI Conference on Web and Social
Media, Vol. 13, 2019, pp. 573–582.
61. A. Pathak, R.K. Srihari, BREAKING! Presenting fake news corpus for automated fact
checking, in: Proceedings of the 57th Annual Meeting of the Association for Computational
Linguistics: Student Research Workshop, 2019, pp. 357–362.
62. Thorne J., Vlachos A., Christodoulopoulos C., Mittal A. 2018. Fever: A large-scale dataset for
fact extraction and verification. arXiv preprint arXiv:1803.05355. [Google Scholar]
63. Shahi G.K., Nandini D. 2020. FakeCovid–A multilingual cross-domain fact check news dataset
for COVID-19. arXiv preprint arXiv:2006.11343. [Google Scholar]
64. T. Mitra, E. Gilbert, Credbank: A large-scale social media corpus with associated credibility
annotations, in: Ninth International AAAI Conference on Web and Social Media, 2015.
65. J. Leskovec, L. Backstrom, J. Kleinberg, Meme-tracking and the dynamics of the news cycle,
in: Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and
Data Mining, 2009, pp. 497–506.
66. G.C. Santia, J.R. Williams, Buzzface: A news veracity dataset with Facebook user
commentary and egos, in: Twelfth International AAAI Conference on Web and Social Media,
2018.
67. Tacchini E., Ballarin G., Della Vedova M.L., Moret S., de Alfaro L. 2017. Some like it hoax:
Automated fake news detection in social networks. arXiv preprint arXiv:1704.07506. [Google
Scholar]
68. De Domenico M., Lima A., Mougel P., Musolesi M. The anatomy of a scientific rumor. Sci.
Rep. 2013;3:2980. [PMC free article] [PubMed] [Google Scholar]
69. Khan T., Michalas A. Trust and believe-should we? Evaluating the trustworthiness of Twitter
users. 2020 IEEE 19th International Conference on Trust, Security and Privacy in Computing and
Communications; TrustCom; IEEE; 2020. pp. 1791–1800. [Google Scholar]
70. Barbado R., Araque O., Iglesias C.A. A framework for fake review detection in online
consumer electronics retailers. Inf. Process. Manage. 2019;56(4):1234–1244. [Google Scholar]
71. Zubiaga A., Aker A., Bontcheva K., Liakata M., Procter R. 2017. Detection and resolution of
rumours in social media: A survey. Preprint, arXiv 1704. [Google Scholar]
72. A. Vlachos, S. Riedel, Fact checking: Task definition and dataset construction, in:
Proceedings of the ACL 2014 Workshop on Language Technologies and Computational Social
Science, 2014, pp. 18–22.
73. W. Ferreira, A. Vlachos, Emergent: A novel data-set for stance classification, in: Proceedings
of the 2016 Conference of the North American Chapter of the Association for Computational
Linguistics: Human Language Technologies, 2016, pp. 1163–1168.
74. Horne B., Adali S. vol. 11. 2017. This just in: Fake news packs a lot in title, uses simpler,
repetitive content in text body, more similar to satire than real news. (Proceedings of the
International AAAI Conference on Web and Social Media). [Google Scholar]
75. C. Burfoot, T. Baldwin, Automatic satire detection: Are you having a laugh?, in: Proceedings
of the ACL-IJCNLP 2009 Conference Short Papers, 2009, pp. 161–164.
76. Torabi Asr F., Taboada M. Big data and quality data for fake news and misinformation
detection. Big Data Soc. 2019;6(1) [Google Scholar]
77. Ott M., Choi Y., Cardie C., Hancock J.T. 2011. Finding deceptive opinion spam by any stretch
of the imagination. arXiv preprint arXiv:1107.4557. [Google Scholar]
78. Riedel B., Augenstein I., Spithourakis G.P., Riedel S. 2017. A simple but tough-to-beat
baseline for the fake news challenge stance detection task. arXiv
preprint arXiv:1707.03264. [Google Scholar]
79. Dutta P.S., Das M., Biswas S., Bora M., Saikia S.S. Fake news prediction: A survey. Int. J. Sci.
Eng. Sci. 2019;3(3):1–3. [Google Scholar]
80. H. Rashkin, E. Choi, J.Y. Jang, S. Volkova, Y. Choi, Truth of varying shades: Analyzing language
in fake news and political fact-checking, in: Proceedings of the 2017 Conference on Empirical
Methods in Natural Language Processing, 2017, pp. 2931–2937.
81. Barrón-Cedeno A., Jaradat I., Da San Martino G., Nakov P. Proppy: Organizing the news
based on their propagandistic content. Inf. Process. Manage. 2019;56(5):1849–1864. [Google
Scholar]
82. Nørregaard J., Horne B.D., Adalı S. vol. 13. 2019. Nela-gt-2018: A large multi-labelled news
dataset for the study of misinformation in news articles; pp. 630–638. (Proceedings of the
International AAAI Conference on Web and Social Media). [Google Scholar]
83. Jang Y., Park C.-H., Seo Y.-S. Fake news analysis modeling using quote
retweet. Electronics. 2019;8(12):1377. [Google Scholar]
84. Papadopoulou O., Zampoglou M., Papadopoulos S., Kompatsiaris I. A corpus of debunked
and verified user-generated videos. Online Inf. Rev. 2019 [Google Scholar]
85. Boididou C., Papadopoulos S., Zampoglou M., Apostolidis L., Papadopoulou O., Kompatsiaris
Y. Detection and visualization of misleading content on Twitter. Int. J. Multimedia Inf.
Retr. 2018;7(1):71–86. [Google Scholar]
86. Jwa H., Oh D., Park K., Kang J.M., Lim H. Exbake: Automatic fake news detection model
based on bidirectional encoder representations from transformers (bert) Appl.
Sci. 2019;9(19):4062. [Google Scholar]
87. Tam N.T., Weidlich M., Zheng B., Yin H., Hung N.Q.V., Stantic B. From anomaly detection to
rumour detection using data streams of social platforms. Proc. VLDB Endow. 2019;12(9):1016–
1029. [Google Scholar]
88. Dai E., Sun Y., Wang S. vol. 14. 2020. Ginger cannot cure cancer: Battling fake health news
with a comprehensive data repository; pp. 853–862. (Proceedings of the International AAAI
Conference on Web and Social Media). [Google Scholar]
89. Hill J.A., Agewall S., Baranchuk A., Booz G.W., Borer J.S., Camici P.G., Chen P.-S., Dominiczak
A.F., Erol Ç., Grines C.L., et al. 2019. Medical misinformation: Vet the message! [PMC free
article] [PubMed] [Google Scholar]
90. Potthast M., Köpsel S., Stein B., Hagen M. European Conference on Information
Retrieval. Springer; 2016. Clickbait detection; pp. 810–817. [Google Scholar]
91. V.L. Rubin, N. Conroy, Y. Chen, S. Cornwell, Fake news or truth? Using satirical cues to
detect potentially misleading news, in: Proceedings of the Second Workshop on Computational
Approaches To Deception Detection, 2016, pp. 7–17.
92. Vosoughi S., Mohsenvand M.N., Roy D. Rumor gauge: Predicting the veracity of rumors on
Twitter. ACM Trans. Knowl. Discov. Data (TKDD) 2017;11(4):1–36. [Google Scholar]
93. Qin Y., Wurzer D., Lavrenko V., Tang C. 2016. Spotting rumors via novelty detection. arXiv
preprint arXiv:1611.06322. [Google Scholar]
94. V. Qazvinian, E. Rosengren, D. Radev, Q. Mei, Rumor has it: Identifying misinformation in
microblogs, in: Proceedings of the 2011 Conference on Empirical Methods in Natural Language
Processing, 2011, pp. 1589–1599.
95. Zubiaga A., Liakata M., Procter R. 2016. Learning reporting dynamics during breaking news
for rumour detection in social media. arXiv preprint arXiv:1610.07363. [Google Scholar]
96. Briscoe E.J., Appling D.S., Hayes H. 2014 47th Hawaii International Conference on System
Sciences. IEEE; 2014. Cues to deception in social media communications; pp. 1435–
1443. [Google Scholar]
97. Chua A.Y., Banerjee S. vol. 1. 2016. Linguistic predictors of rumor veracity on the internet;
pp. 387–391. (Proceedings of the International MultiConference of Engineers and Computer
Scientists). [Google Scholar]
98. J. Ito, J. Song, H. Toda, Y. Koike, S. Oyama, Assessment of tweet credibility with LDA
features, in: Proceedings of the 24th International Conference on World Wide Web, 2015, pp.
953–958.
99. Ma J., Gao W., Mitra P., Kwon S., Jansen B.J., Wong K.-F., Cha M. 2016. Detecting rumors
from microblogs with recurrent neural networks. [Google Scholar]
100. Jin Z., Cao J., Zhang Y., Zhou J., Tian Q. Novel visual and statistical image features for
microblogs news verification. IEEE Trans. Multimed. 2016;19(3):598–608. [Google Scholar]
101. Potthast M., Kiesel J., Reinartz K., Bevendorff J., Stein B. 2017. A stylometric inquiry into
hyperpartisan and fake news. arXiv preprint arXiv:1702.05638. [Google Scholar]
102. C. Castillo, M. Mendoza, B. Poblete, Information credibility on twitter, in: Proceedings of
the 20th International Conference on World Wide Web, 2011, pp. 675–684.
103. Hamidian S., Diab M.T. 2019. Rumor detection and classification for Twitter data. arXiv
preprint arXiv:1912.08926. [Google Scholar]
104. Hu X., Tang J., Liu H. vol. 28. 2014. Online social spammer detection. (Proceedings of the
AAAI Conference on Artificial Intelligence). [Google Scholar]
105. Kwon S., Cha M., Jung K., Chen W., Wang Y. 2013 IEEE 13th International Conference on
Data Mining. IEEE; 2013. Prominent features of rumor propagation in online social media; pp.
1103–1108. [Google Scholar]
106. Jin Z., Cao J., Zhang Y., Luo J. vol. 30. 2016. News verification by exploiting conflicting social
viewpoints in microblogs. (Proceedings of the AAAI Conference on Artificial
Intelligence). [Google Scholar]
107. Bondielli A., Marcelloni F. A survey on fake news and rumour detection
techniques. Inform. Sci. 2019;497:38–55. [Google Scholar]
108. Silva A., Luo L., Karunasekera S., Leckie C. vol. 35. 2021. Embracing domain differences in
fake news: Cross-domain fake news detection using multi-modal data; pp. 557–565.
(Proceedings of the AAAI Conference on Artificial Intelligence). [Google Scholar]
109. Hoens T.R., Polikar R., Chawla N.V. Learning from streaming data with concept drift and
imbalance: An overview. Progress Artif. Intell. 2012;1(1):89–101. [Google Scholar]
110. Devlin J., Chang M.-W., Lee K., Toutanova K. 2018. Bert: Pre-training of deep bidirectional
transformers for language understanding. arXiv preprint arXiv:1810.04805. [Google Scholar]
111. Peters M.E., Neumann M., Iyyer M., Gardner M., Clark C., Lee K., Zettlemoyer L. 2018.
Deep contextualized word representations. CoRR abs/1802.05365, arXiv
preprint arXiv:1802.05365, 1802. [Google Scholar]
112. Mikolov T., Grave E., Bojanowski P., Puhrsch C., Joulin A. 2017. Advances in pre-training
distributed word representations. arXiv preprint arXiv:1712.09405. [Google Scholar]
113. Grave E., Bojanowski P., Gupta P., Joulin A., Mikolov T. 2018. Learning word vectors for 157
languages. arXiv preprint arXiv:1802.06893. [Google Scholar]
114. J. Pennington, R. Socher, C.D. Manning, Glove: Global vectors for word representation, in:
Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing,
EMNLP, 2014, pp. 1532–1543.
115. Koloski B., Perdih T.S., Robnik-Šikonja M., Pollak S., Škrlj B. Knowledge graph informed fake
news classification via heterogeneous representation
ensembles. Neurocomputing. 2022 [Google Scholar]
116. Sun Z., Deng Z.-H., Nie J.-Y., Tang J. 2019. Rotate: Knowledge graph embedding by
relational rotation in complex space. arXiv preprint arXiv:1902.10197. [Google Scholar]
117. Zhang S., Tay Y., Yao L., Liu Q. Quaternion knowledge graph embeddings. Adv. Neural Inf.
Process. Syst. 2019;32 [Google Scholar]
118. Trouillon T., Welbl J., Riedel S., Gaussier É., Bouchard G. International Conference on
Machine Learning. PMLR; 2016. Complex embeddings for simple link prediction; pp. 2071–
2080. [Google Scholar]
119. Pérez-Rosas V., Kleinberg B., Lefevre A., Mihalcea R. 2017. Automatic detection of fake
news. arXiv preprint arXiv:1708.07104. [Google Scholar]
120. Cho K., Van Merriënboer B., Gulcehre C., Bahdanau D., Bougares F., Schwenk H., Bengio Y.
2014. Learning phrase representations using RNN encoder-decoder for statistical machine
translation. arXiv preprint arXiv:1406.1078. [Google Scholar]
121. Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A.N., Kaiser Ł., Polosukhin
I. Attention is all you need. Adv. Neural Inf. Process. Syst. 2017;30 [Google Scholar]
122. Glazkova A., Glazkov M., Trifonov T. Springer; 2021. g2tmn at Constraint @Aaai2021:
Exploiting CT-BERT and Ensembling Learning for COVID-19 Fake News Detection; pp. 116–
127. [Google Scholar]
123. A. Li, Z. Qin, R. Liu, Y. Yang, D. Li, Spam review detection with graph convolutional
networks, in: Proceedings of the 28th ACM International Conference on Information and
Knowledge Management, 2019, pp. 2703–2711.
124. Lin H., Zhang X., Fu X. A graph convolutional encoder and decoder model for rumor
detection. 2020 IEEE 7th International Conference on Data Science and Advanced Analytics;
DSAA; IEEE; 2020. pp. 300–306. [Google Scholar]
125. Vu D.T., Jung J.J. Rumor detection by propagation embedding based on graph
convolutional network. Int. J. Comput. Intell. Syst. 2021;14(1):1053–1065. [Google Scholar]
126. Simonyan K., Zisserman A. 2014. Very deep convolutional networks for large-scale image
recognition. arXiv preprint arXiv:1409.1556. [Google Scholar]
127. Nickel M., Murphy K., Tresp V., Gabrilovich E. A review of relational machine learning for
knowledge graphs. Proc. IEEE. 2015;104(1):11–33. [Google Scholar]
128. Ginsca A.L., Popescu A., Lupu M. Credibility in information retrieval. Found. Trends Inf.
Retr. 2015;9(5):355–475. [Google Scholar]
129. Etzioni O., Banko M., Soderland S., Weld D.S. Open information extraction from the
web. Commun. ACM. 2008;51(12):68–74. [Google Scholar]
130. A. Magdy, N. Wanas, Web-based statistical fact checking of textual documents, in:
Proceedings of the 2nd International Workshop on Search and Mining User-Generated
Contents, 2010, pp. 103–110.
131. De Alfaro L., Polychronopoulos V., Shavlovsky M. vol. 3. 2015. Reliable aggregation of
Boolean crowdsourced tasks. (Proceedings of the AAAI Conference on Human Computation and
Crowdsourcing). [Google Scholar]
132. Chen Y., Conroy N.K., Rubin V.L. News in an online world: The need for an “automatic crap
detector” Proc. Assoc. Inf. Sci. Technol. 2015;52(1):1–4. [Google Scholar]
133. V.L. Rubin, N.J. Conroy, Y. Chen, Towards news verification: Deception detection methods
for news discourse, in: Hawaii International Conference on System Sciences, 2015, pp. 5–8.
134. Wei Z., Chen J., Gao W., Li B., Zhou L., He Y., Wong K.-F. Social Media Content Analysis:
Natural Language Processing and beyond. World Scientific; 2018. An empirical study on
uncertainty identification in social media context; pp. 79–88. [Google Scholar]
135. B. Shi, T. Weninger, Fact checking in heterogeneous information networks, in: Proceedings
of the 25th International Conference Companion on World Wide Web, 2016, pp. 101–102.
136. Shu K., Mahudeswaran D., Wang S., Lee D., Liu H. 2018. Fakenewsnet: A data repository
with news content, social context and spatialtemporal information for studying fake news on
social media. arXiv preprint arXiv:1809.01286. [PubMed] [Google Scholar]
137. Sitaula N., Mohan C.K., Grygiel J., Zhou X., Zafarani R. Disinformation, Misinformation, and
Fake News in Social Media. Springer; 2020. Credibility-based fake news detection; pp. 163–
182. [Google Scholar]
138. Mocanu D., Rossi L., Zhang Q., Karsai M., Quattrociocchi W. Collective attention in the age
of (mis) information. Comput. Hum. Behav. 2015;51:1198–1204. [Google Scholar]
139. M. Tambuscio, G. Ruffo, A. Flammini, F. Menczer, Fact-checking effect on viral hoaxes: A
model of misinformation spread in social networks, in: Proceedings of the 24th International
Conference on World Wide Web, 2015, pp. 977–982.
140. Ma J., Gao W., Wong K.-F. Association for Computational Linguistics; 2017. Detect Rumors
in Microblog Posts Using Propagation Structure Via Kernel Learning. [Google Scholar]
141. L. Wu, H. Liu, Tracing fake-news footprints: Characterizing social media messages by how
they propagate, in: Proceedings of the Eleventh ACM International Conference on Web Search
and Data Mining, 2018, pp. 637–645.
142. Y. Liu, Y.-F.B. Wu, Early detection of fake news on social media through propagation path
classification with recurrent and convolutional networks, in: Thirty-Second AAAI Conference on
Artificial Intelligence, 2018.
143. Shu K., Mahudeswaran D., Wang S., Liu H. vol. 14. 2020. Hierarchical propagation networks
for fake news detection: Investigation and exploitation; pp. 626–637. (Proceedings of the
International AAAI Conference on Web and Social Media). [Google Scholar]
144. Wu K., Yang S., Zhu K.Q. 2015 IEEE 31st International Conference on Data
Engineering. IEEE; 2015. False rumors detection on sina weibo by propagation structures; pp.
651–662. [Google Scholar]
145. Zhou X., Zafarani R. Network-based fake news detection: A pattern-driven approach. ACM
SIGKDD Explor. Newsl. 2019;21(2):48–60. [Google Scholar]
146. Alonso-Bartolome S., Segura-Bedmar I. 2021. Multimodal fake news detection. arXiv
preprint arXiv:2112.04831. [Google Scholar]
147. Malhotra B., Vishwakarma D.K. Classification of propagation path and tweets for rumor
detection using graphical convolutional networks and transformer based encodings. 2020 IEEE
Sixth International Conference on Multimedia Big Data; BigMM; IEEE; 2020. pp. 183–
190. [Google Scholar]
148. Zhou X., Wu J., Zafarani R. Pacific-Asia Conference on Knowledge Discovery and Data
Mining. Springer; 2020. SAFE: Similarity-aware multi-modal fake news detection; pp. 354–
367. [Google Scholar]
149. Shang L., Zhang Y., Zhang D., Wang D. Fauxward: A graph neural network approach to
fauxtography detection using social media comments. Soc. Netw. Anal. Min. 2020;10(1):1–
16. [Google Scholar]
150. Paraschiv A., Zaharia G.-E., Cercel D.-C., Dascalu M. Graph convolutional networks applied
to fakenews: Corona virus and 5G conspiracy. Univ. Politeh. Bucharest Sci. Bull. Ser. C-Electr.
Eng. Comput. Sci. 2021:71–82. [Google Scholar]
151. Zhang L., Li J., Zhou B., Jia Y. Rumor detection based on SAGNN: Simplified aggregation
graph neural networks. Mach. Learn. Knowl. Extr. 2021;3(1):84–94. [Google Scholar]
152. Ma J., Gao W., Wong K.-F. Association for Computational Linguistics; 2018. Rumor
Detection on Twitter with Tree-Structured Recursive Neural Networks. [Google Scholar]
153. Meyers M., Weiss G., Spanakis G. Multidisciplinary International Symposium on
Disinformation in Open Online Media. Springer; 2020. Fake news detection on Twitter using
propagation structures; pp. 138–158. [Google Scholar]
154. Bai N., Meng F., Rui X., Wang Z. Rumour detection based on graph convolutional neural
net. IEEE Access. 2021;9:21686–21693. [Google Scholar]
155. N. Tuan, P. Minh, FakeNews detection using pre-trained language models and graph
convolutional networks, in: Multimedia Evaluation Benchmark Workshop 2020, MediaEval
2020, 2020.
156. Ke Z., Li Z., Zhou C., Sheng J., Silamu W., Guo Q. Rumor detection on social media via fused
semantic information and a propagation heterogeneous
graph. Symmetry. 2020;12(11):1806. [Google Scholar]
157. Horne B.D., Nørregaard J., Adalı S. vol. 13. 2019. Different spirals of sameness: A study of
content sharing in mainstream and alternative media; pp. 257–266. (Proceedings of the
International AAAI Conference on Web and Social Media). [Google Scholar]
158. Bachi G., Coscia M., Monreale A., Giannotti F. 2012 International Conference on Privacy,
Security, Risk and Trust and 2012 International Confernece on Social Computing. IEEE; 2012.
Classifying trust/distrust relationships in online social networks; pp. 552–557. [Google Scholar]
159. G. Wang, X. Zhang, S. Tang, H. Zheng, B.Y. Zhao, Unsupervised clickstream clustering for
user behavior analysis, in: Proceedings of the 2016 CHI Conference on Human Factors in
Computing Systems, 2016, pp. 225–236.
160. Hamilton W.L. Graph representation learning. Synthesis Lect. Artif. Intell. Mach.
Learn. 2020;14(3):1–159. [Google Scholar]
161. Bruna J., Zaremba W., Szlam A., LeCun Y. 2013. Spectral networks and locally connected
networks on graphs. arXiv preprint arXiv:1312.6203. [Google Scholar]
162. Niepert M., Ahmed M., Kutzkov K. International Conference on Machine Learning. PMLR;
2016. Learning convolutional neural networks for graphs; pp. 2014–2023. [Google Scholar]
163. Ying R., You J., Morris C., Ren X., Hamilton W.L., Leskovec J. 2018. Hierarchical graph
representation learning with differentiable pooling. arXiv preprint arXiv:1806.08804. [Google
Scholar]
164. Kipf T.N., Welling M. 2016. Semi-supervised classification with graph convolutional
networks. arXiv preprint arXiv:1609.02907. [Google Scholar]
165. Jiang X., Zhu R., Li S., Ji P. Co-embedding of nodes and edges with graph neural
networks. IEEE Trans. Pattern Anal. Mach. Intell. 2020 [PubMed] [Google Scholar]
166. Sperduti A., Starita A. Supervised neural networks for the classification of structures. IEEE
Trans. Neural Netw. 1997;8(3):714–735. [PubMed] [Google Scholar]
167. Gori M., Monfardini G., Scarselli F. Proceedings. 2005 IEEE International Joint Conference
on Neural Networks, 2005, Vol. 2. IEEE; 2005. A new model for learning in graph domains; pp.
729–734. [Google Scholar]
168. Scarselli F., Gori M., Tsoi A.C., Hagenbuchner M., Monfardini G. The graph neural network
model. IEEE Trans. Neural Netw. 2008;20(1):61–80. [PubMed] [Google Scholar]
169. Gallicchio C., Micheli A. Graph echo state networks. The 2010 International Joint
Conference on Neural Networks; IJCNN; IEEE; 2010. pp. 1–8. [Google Scholar]
170. Kipf T.N., Welling M. 2016. Variational graph auto-encoders. arXiv
preprint arXiv:1611.07308. [Google Scholar]
171. Y. Wang, B. Xu, M. Kwak, X. Zeng, A simple training strategy for graph autoencoder, in:
Proceedings of the 2020 12th International Conference on Machine Learning and Computing,
2020, pp. 341–345.
172. Li Y., Yu R., Shahabi C., Liu Y. 2017. Diffusion convolutional recurrent neural network: Data-
driven traffic forecasting. arXiv preprint arXiv:1707.01926. [Google Scholar]
173. Seo Y., Defferrard M., Vandergheynst P., Bresson X. International Conference on Neural
Information Processing. Springer; 2018. Structured sequence modeling with graph
convolutional recurrent networks; pp. 362–373. [Google Scholar]
174. Zhang J., Shi X., Xie J., Ma H., King I., Yeung D.-Y. 2018. Gaan: Gated attention networks for
learning on large and spatiotemporal graphs. arXiv preprint arXiv:1803.07294. [Google Scholar]
175. Wu Z., Pan S., Long G., Jiang J., Zhang C. 2019. Graph wavenet for deep spatial-temporal
graph modeling. arXiv preprint arXiv:1906.00121. [Google Scholar]
176. S. Yan, Y. Xiong, D. Lin, Spatial temporal graph convolutional networks for skeleton-based
action recognition, in: Thirty-Second AAAI Conference on Artificial Intelligence, 2018.
177. Yu B., Yin H., Zhu Z. 2017. Spatio-temporal graph convolutional networks: A deep learning
framework for traffic forecasting. arXiv preprint arXiv:1709.04875. [Google Scholar]
178. Veličković P., Cucurull G., Casanova A., Romero A., Lio P., Bengio Y. 2017. Graph attention
networks. arXiv preprint arXiv:1710.10903. [Google Scholar]
179. Thekumparampil K.K., Wang C., Oh S., Li L.-J. 2018. Attention-based graph neural network
for semi-supervised learning. arXiv preprint arXiv:1803.03735. [Google Scholar]
180. J.B. Lee, R. Rossi, X. Kong, Graph classification using structural attention, in: Proceedings of
the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 2018,
pp. 1666–1674.
181. Liberati A., Altman D.G., Tetzlaff J., Mulrow C., Gøtzsche P.C., Ioannidis J.P., Clarke M.,
Devereaux P.J., Kleijnen J., Moher D. The PRISMA statement for reporting systematic reviews
and meta-analyses of studies that evaluate health care interventions: Explanation and
elaboration. J. Clin. Epidemiol. 2009;62(10):e1–e34. [PubMed] [Google Scholar]
182. Afroz S., Brennan M., Greenstadt R. 2012 IEEE Symposium on Security and Privacy. IEEE;
2012. Detecting hoaxes, frauds, and deception in writing style online; pp. 461–475. [Google
Scholar]
183. Lafferty J., McCallum A., Pereira F.C. 2001. Conditional random fields: Probabilistic models
for segmenting and labeling sequence data. [Google Scholar]
184. Vosoughi S. Massachusetts Institute of Technology; 2015. Automatic Detection and
Verification of Rumors on Twitter. (Ph.D. thesis) [Google Scholar]
185. O. Ajao, D. Bhowmik, S. Zargari, Fake news identification on twitter with hybrid CNN and
RNN models, in: Proceedings of the 9th International Conference on Social Media and Society,
2018, pp. 226–230.
186. Jacovi A., Shalom O.S., Goldberg Y. 2018. Understanding convolutional neural networks for
text classification. arXiv preprint arXiv:1809.08037. [Google Scholar]
187. S. Volkova, K. Shaffer, J.Y. Jang, N. Hodas, Separating facts from fiction: Linguistic models
to classify suspicious and trusted news posts on twitter, in: Proceedings of the 55th Annual
Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), 2017, pp.
647–653.
188. F. Yu, Q. Liu, S. Wu, L. Wang, T. Tan, et al., A Convolutional Approach for Misinformation
Identification, in: IJCAI, 2017, pp. 3901–3907.
189. Zubiaga A., Liakata M., Procter R., Wong Sak Hoi G., Tolmie P. Analysing how people orient
to and spread rumours in social media by looking at conversational threads. PLoS
One. 2016;11(3) [PMC free article] [PubMed] [Google Scholar]
190. Hu G., Ding Y., Qi S., Wang X., Liao Q. CCF International Conference on Natural Language
Processing and Chinese Computing. Springer; 2019. Multi-depth graph convolutional networks
for fake news detection; pp. 698–710. [Google Scholar]
191. C. Li, D. Goldwasser, Encoding social information with graph convolutional networks
forpolitical perspective detection in news media, in: Proceedings of the 57th Annual Meeting of
the Association for Computational Linguistics, 2019, pp. 2594–2604.
192. Benamira A., Devillers B., Lesot E., Ray A.K., Saadi M., Malliaros F.D. Semi-supervised
learning and graph neural networks for fake news detection. 2019 IEEE/ACM International
Conference on Advances in Social Networks Analysis and Mining; ASONAM; IEEE; 2019. pp.
568–569. [Google Scholar]
193. Hamid A., Shiekh N., Said N., Ahmad K., Gul A., Hassan L., Al-Fuqaha A. 2020. Fake news
detection in social media using graph neural networks and nlp techniques: A COVID-19 use-
case. arXiv preprint arXiv:2012.07517. [Google Scholar]
194. Z. Pehlivan, On the pursuit of fake news: From graph convolutional networks to time
series, in: Multimedia Evaluation Benchmark Workshop 2020, MediaEval 2020, 2020.
195. G.-A. Vlad, G.-E. Zaharia, D.-C. Cercel, M. Dascalu, UPB@ DANKMEMES: Italian Memes
Analysis-Employing Visual Models and Graph Convolutional Networks for Meme Identification
and Hate Speech Detection (short paper), in: EVALITA, 2020.
196. M. Dong, B. Zheng, N. Quoc Viet Hung, H. Su, G. Li, Multiple rumor source detection with
graph convolutional networks, in: Proceedings of the 28th ACM International Conference on
Information and Knowledge Management, 2019, pp. 569–578.
197. Zachary W.W. An information flow model for conflict and fission in small groups. J.
Anthropol. Res. 1977;33(4):452–473. [Google Scholar]
198. Lusseau D., Schneider K., Boisseau O.J., Haase P., Slooten E., Dawson S.M. The bottlenose
dolphin community of doubtful sound features a large proportion of long-lasting
associations. Behav. Ecol. Sociobiol. 2003;54(4):396–405. [Google Scholar]
199. Watts D.J., Strogatz S.H. Collective dynamics of ‘small-world’
networks. Nature. 1998;393(6684):440–442. [PubMed] [Google Scholar]
200. Gleiser P.M., Danon L. Community structure in jazz. Adv. Complex Syst. 2003;6(04):565–
573. [Google Scholar]
201. Wu Y., Lian D., Xu Y., Wu L., Chen E. vol. 34. 2020. Graph convolutional networks with
markov random field reasoning for social spammer detection; pp. 1054–1061. (Proceedings of
the AAAI Conference on Artificial Intelligence). [Google Scholar]
202. K. Lee, B.D. Eoff, J. Caverlee, Seven months with the devils: A long-term study of content
polluters on twitter, in: Fifth International AAAI Conference on Weblogs and Social Media, 2011.
203. C. Yang, R. Harkreader, J. Zhang, S. Shin, G. Gu, Analyzing spammers’ social networks for
fun and profit: A case study of cyber criminal ecosystem on Twitter, in: Proceedings of the 21st
International Conference on World Wide Web, 2012, pp. 71–80.
204. Lotfi S., Mirzarezaee M., Hosseinzadeh M., Seydi V. Detection of rumor conversations in
Twitter using graph convolutional networks. Appl. Intell. 2021;51(7):4774–4787. [Google
Scholar]
205. Data61 C. 2018. Stellargraph machine learning library. Publication Title: GitHub Repository.
GitHub. [Google Scholar]
206. Vickers N.J. Animal communication: When i’m calling you, will you answer too? Curr.
Biol. 2017;27(14):R713–R715. [PubMed] [Google Scholar]
207. Y. Liu, M. Ott, N. Goyal, J. Du, M. Joshi, D. Chen, O. Levy, M. Lewis, L. Zettlemoyer, V.
Stoyanov, Roberta: A robustly optimized bert pretraining approach, 2019, arXiv preprint.
208. Pires T., Schlinger E., Garrette D. 2019. How multilingual is multilingual BERT? arXiv
preprint arXiv:1906.01502. [Google Scholar]
209. G. Klambauer, T. Unterthiner, A. Mayr, S. Hochreiter, Self-normalizing neural networks, in:
Proceedings of the 31st International Conference on Neural Information Processing Systems,
2017, pp. 972–981.
210. Zhang Y., Wallace B. 2015. A sensitivity analysis of (and practitioners’ guide to)
convolutional neural networks for sentence classification. arXiv
preprint arXiv:1510.03820. [Google Scholar]
211. W.L. Hamilton, R. Ying, J. Leskovec, Inductive representation learning on large graphs, in:
Proceedings of the 31st International Conference on Neural Information Processing Systems,
2017, pp. 1025–1035.
212. Q. Xu, F. Shen, L. Liu, H.T. Shen, Graphcar: Content-aware multimedia recommendation
with graph autoencoder, in: The 41st International ACM SIGIR Conference on Research &
Development in Information Retrieval, 2018, pp. 981–984.
213. Berg R.v.d., Kipf T.N., Welling M. 2017. Graph convolutional matrix completion. arXiv
preprint arXiv:1706.02263. [Google Scholar]
214. García-Ull F.J. DeepFakes: The next challenge in fake news detection. Anàlisi: Quaderns de
Comunicació I Cultura. 2021;(64):0103–120. [Google Scholar]
215. K. Shu, S. Wang, H. Liu, Beyond news contents: The role of social context for fake news
detection, in: Proceedings of the Twelfth ACM International Conference on Web Search and
Data Mining, 2019, pp. 312–320.
216. Von Luxburg U. A tutorial on spectral clustering. Stat. Comput. 2007;17(4):395–
416. [Google Scholar]
217. Zhou J., Huang J.X., Hu Q.V., He L. SK-GCN: Modeling syntax and knowledge via graph
convolutional network for aspect-level sentiment classification. Knowl.-Based
Syst. 2020;205 [Google Scholar]
218. Yao L., Mao C., Luo Y. vol. 33. 2019. Graph convolutional networks for text classification;
pp. 7370–7377. (Proceedings of the AAAI Conference on Artificial Intelligence). [Google Scholar]
219. Bijari K., Zare H., Kebriaei E., Veisi H. Leveraging deep graph-based text representation for
sentiment polarity applications. Expert Syst. Appl. 2020;144 [Google Scholar]
220. M. Zhang, T. Qian, Convolution over Hierarchical Syntactic and Lexical Graphs for Aspect
Level Sentiment Analysis, in: Proceedings of the 2020 Conference on Empirical Methods in
Natural Language Processing, EMNLP, 2020, pp. 3540–3549.
221. Wissner-Gross A. Datasets over algorithms. Edge. Com. Retr. 2016;8 [Google Scholar]