Fake news, disinformation and misinformation in social media: a review

Classified as probably not a substantial claim.

Even the best AI for detecting fake news remains ineffective.

“Similarly, other researchers concurred that even the best AI for spotting fake news is still ineffective.18”

• cites [18] Avram M, Micallef N, Patil S, Menczer F (2020) Exposure to social engagement metrics increases vulnerability to misinformation. arXiv preprint arxiv:2005.04682, https://doi.org/10.37016/mr-2020-033 sourceCould not verify

  The Implications section describes technical approaches to detecting misinformation, including supervised models and natural-language processing, but does not assess whether even the best AI is ineffective. Searches for AI detection effectiveness and limitations found no passage making that claim; the paper instead focuses on social engagement metrics. (LLM, conf=0.72)

  Experimental — accuracy not yet measured

May combine two claims. The quote only partly states this claim.

Some researchers treat fake news as a type of misinformation, others as a type of disinformation, others as both, and some distinguish it from both.

“Some consider fake news as a type of misinformation (Allen et al. 2020; Singh et al. 2021; Ha et al. 2021; Pennycook and Rand 2019; Shao et al. 2018b; Di Domenico et al. 2021; Sharma et al. 2019; Celliers and Hattingh 2020; Klein and Wueller 2017; Potthast et al. 2017; Islam et al. 2020), others consider it as a type of disinformation”

• cites Allen et al. 2020 sourceSupported
• cites Singh et al. 2021 sourceNot checked — no full text of the source
• cites Ha et al. 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

May combine two claims.

Fact-checking organizations cannot match the speed of propagation because they need human verification, which prevents a timely and cost-effective response.

“Moreover, fact-checking organizations cannot keep up with the dynamics of propagation as they require human verification, which can hold back a timely and cost-effective response (Kim et al. 2018; Ruchansky et al. 2017; Shu et al. 2018a).”

• cites Kim et al. 2018 sourcePartly supported

  The Introduction (p. 1) says fact-checking is costly and that independent reviewers can check only a small percentage of stories spreading over time, supporting the claim about limited capacity and cost. It does not say that human verification causes delays or that organizations cannot keep pace with propagation; the model instead assumes instantaneous verification once a story is sent for checking (Problem formulation, p. 3). (LLM, conf=0.85)

  Experimental — accuracy not yet measured
• cites Ruchansky et al. 2017 sourcePartly supported

  The Introduction (p. 1) says fact-checking websites require expert analysis, which inhibits a timely response. It does not establish that they cannot keep up with the speed of propagation or that their response is not cost-effective. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites Shu et al. 2018aNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Most existing reviews omit at least one perspective and often do not cover detection approaches using blockchain and fact-checking, or analyses of search engine optimization metrics.

“Furthermore, most of these studies ignore at least one of the mentioned perspectives, and in many cases, they do not cover other existing detection approaches using methods such as blockchain and fact-checking, as well as analysis on metrics used for Search Engine Optimization (Mazzeo and Rapisarda 2022).”

• cites Mazzeo and Rapisarda 2022 sourcePartly supported

  The Introduction (page unavailable) says SEO metrics have not been widely used to identify fake-news sources, supporting the SEO portion of the claim. The cited passages do not establish that most existing reviews omit a perspective or fail to cover blockchain and fact-checking approaches; the paper itself describes using fact-checking sources for data labeling. (LLM, conf=0.77)

  Experimental — accuracy not yet measured

May combine two claims.

Fake news can strongly affect society because manipulated false content is easier to create and harder to detect.

“Hence, fake news can have a significant impact on society as manipulated and false content is easier to generate and harder to detect (Kumar and Shah 2018)”

• cites Kumar and Shah 2018 sourcePartly supported

  The introduction (p. 2) and Section 4 (p. 10) describe substantial social impacts of false information. Section 6 (p. 21) says false information is made to appear true and is therefore harder to identify. The discussion on p. 8 says it can be generated automatically, but does not establish the claim that manipulated false content is easier to create or that this ease of creation explains its societal impact. (LLM, conf=0.83)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

The first definition of fake news was given by Allcott and Gentzkow (2017): news articles that are intentionally and verifiably false and could mislead readers.

“The first definition of the term fake news was provided by Allcott and Gentzkow (2017) as news articles that are intentionally and verifiably false and could mislead readers.”

• cites Allcott and Gentzkow (2017) sourcePartly supported

  The “Definition and History” section gives the quoted definition almost verbatim. It does not say that Allcott and Gentzkow provided the first definition of the term, so the priority claim is unsupported. Page numbers were not available in our copy. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

May combine two claims.

People with little knowledge of an issue are more easily convinced that misleading or fake news is true than people strongly concerned about fake news, especially when the content is shared as video rather than text or audio.

“For instance, people with little knowledge about the issue compared to those who are strongly concerned about the key issue of fake news tend to be easier to convince that the misleading or fake news is real, especially when shared via a video modality as compared to the text or the audio modality (Demuyakor and Opata 2022).”

• cites Demuyakor and Opata 2022 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Leveraging personality traits and linguistic patterns can improve performance in differentiating between fact-checkers and fake news spreaders.

“They show that leveraging personality traits and linguistic patterns can improve the performance in differentiating between checkers and spreaders.”

• cites Giachanou et al. 2020 sourceSupported

May combine two claims.

Rumors are not necessarily false and may turn out to be true.

“However, they are not necessarily false and may turn out to be true (Zubiaga et al. 2018).”

• cites Zubiaga et al. 2018 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Conspiracy belief is associated with psychological, political, and social factors.

“Conspiracy belief refers to people’s adoption and belief of conspiracy theories, and it is associated with psychological, political and social factors (Douglas et al. 2019).”

• cites Douglas et al. 2019 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

During the COVID-19 pandemic, conspiracy theories have been discussed from a public health perspective.

“For instance, lately and during the COVID-19 pandemic, conspiracy theories have been discussed from a public health perspective (Meese et al. 2020; Allington et al. 2020; Freeman et al. 2020).”

• cites Meese et al. 2020 sourceSupported
• cites Allington et al. 2020 sourceSupported
• cites Freeman et al. 2020 sourceSupported

May combine two claims.

Individuals most likely to engage with fake news sources during and after the 2016 US presidential election were generally conservative-leaning, older, and highly engaged with political news.

“found that individuals most likely to engage with fake news sources were generally conservative-leaning, older, and highly engaged with political news.”

• cites Allcott and Gentzkow 2017 sourceCould not verify

  The Partisanship of Fake News section reports that pro-Trump stories were shared more often, but does not identify the people who engaged with fake-news sources. The Who Believes Fake News section associates older age and greater media consumption with more accurate judgments about headlines, not greater engagement with fake-news sources. Our copy does not establish the claimed demographic profile. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites Grinberg et al. 2019 sourceNot checked — no full text of the source
• cites Guess et al. 2019 sourcePartly supported

  The Results and Discussion (pages not provided) support the findings that older and more conservative respondents were more likely to share links from fake-news domains on Facebook during the 2016 election. They do not establish that these respondents were highly engaged with political news: the study measured sharing rather than engagement with news sources, and the Results say that people who shared the most content overall were, if anything, less likely to share from fake-news domains. The Discussion describes the finding as occurring during the election campaign, not after it. (LLM, conf=0.89)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Content-based fake news forms can include false text such as hyperlinks or embedded content, multimedia such as false videos, images and audios, and multimodal content.

“forms of fake news may include false text such as hyperlinks or embedded content; multimedia such as false videos (Demuyakor and Opata 2022), images (Masciari et al. 2020; Shen et al. 2019), audios (Demuyakor and Opata 2022) and so on. Moreover, we can also find multimodal content (Shu et al. 2020a)”

• cites Demuyakor and Opata 2022 sourceSupported
• cites Masciari et al. 2020 sourceNot checked — no full text of the source
• cites Shen et al. 2019 sourceSupported

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Contextual features fall into two groups, those about users and those about networks, and both can support analysis and model training in AI and ML systems.

“Social context-based aspects can be classified into two subcategories, user-based and network-based, and they can be used for context analysis and training tasks in the case of AI- and ML-based approaches.”

• cites Shu et al. 2019bNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Wang et al. 2019c sourceCould not verify

  Section 3.2 distinguishes individual-level psychological approaches from network-science approaches, but it does not classify social-context features as user-based and network-based. Sections 3.3 and 4.1 discuss content and network analyses, not using those two feature categories for context analysis or AI/ML training. Searches for the feature taxonomy and machine-learning training found no passage addressing the claim; page numbers were unavailable in our copy. (LLM, conf=0.73)

  Experimental — accuracy not yet measured
• cites Hamdi et al. 2020 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Social bots increase public exposure to created content and boost its perceived credibility by targeting influential users more likely to believe disinformation in the hope of getting them to repost the fabricated content.

“Next, they try to increase public exposure to the created content and thus boost its perceived credibility by targeting influential users that are more likely to believe disinformation in the hope of getting them to “repost” the fabricated content.”

• cites Shu et al. 2020a sourcePartly supported

  Section 3.3, “Bots on Social Media,” says bots target people more likely to believe disinformation. It also says bots target influential users in hopes of a repost that increases exposure and perceived credibility. It does not establish that the influential users are the people more likely to believe disinformation. (LLM, conf=0.83)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim. May combine two claims.

Fake news is diverse and dynamic in content, topics, publishing methods, media platforms, and sophisticated linguistic styles geared to emulate true news.

“fake news is diverse and dynamic in terms of content, topics, publishing methods and media platforms, and sophisticated linguistic styles geared to emulate true news.”

• cites Shu et al. 2020dNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Wang et al. 2020 sourcePartly supported

  The Introduction (p. 1) describes fake news about several topics, its spread on social media, and the dynamic nature of news articles. The Insight Analysis (p. 8) also finds that news-content distributions change over time. Our copy does not substantiate diverse publishing methods or sophisticated linguistic styles designed to emulate true news. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites Pathak and Srihari 2019 sourcePartly supported

  Section 4.2 (page not provided) observes that fake-news writers use mainstream writing conventions, supporting the claim that their style can emulate genuine news. Section 3 mentions different dissemination techniques, but the paper focuses on 2016 US election articles; it does not establish the broader claim that fake news is diverse and dynamic across topics, publishing methods, and media platforms. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Techniques used to detect bots depend on the social platform and the objective.

“Ismailov et al. (2020) assert that the techniques used to detect bots depend on the social platform and the objective.”

• cites Ismailov et al. 2020 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Some studies prove that deep learning techniques outperform traditional machine learning techniques.

“Some even prove that deep learning techniques outperform traditional machine learning techniques (Mishra et al. 2022).”

• cites Mishra et al. 2022 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

User behavior models primarily focus on features related to an individual user's behavior, such as frequency of activities, patterns of activity, and clickstream sequences.

“User behavior models primarily focus on features related to an individual user’s behavior, such as frequency of activities (e.g., number of tweets or posts per time interval), patterns of activity and clickstream sequences.”

• cites Ismailov et al. 2020 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Earlier research found that users distrust automated tools for detecting fake news.

“Previous work by Yaqub et al. (2020) has shown that people lack trust in automated solutions for fake news detection”

• cites Yaqub et al. (2020) sourcePartly supported

  The Study Design and Flow section says the study tested a warning attributed to AI algorithms, and Effect of Credibility Indicators reports that this warning was less effective than a fact-checker warning at reducing intent to share non-true headlines. That supports a narrower finding about responses to AI-labelled warnings, but the paper does not establish that users generally distrust automated fake-news detection tools or directly measure such trust. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

The Russia-based Internet Research Agency disseminated inflammatory content online to influence the outcome of the 2016 U.S. presidential election.

“A prominent troll farm example is the Russia-based Internet Research Agency (IRA), which disseminated inflammatory content online to influence the outcome of the 2016 U.S. presidential election.33”

• cites [33] Brashier NM, Pennycook G, Berinsky AJ, Rand DG (2021) Timing matters when correcting fake news. Proc Natl Acad Sci. https://doi.org/10.1073/pnas.2020043118 sourceCould not verify

  The introduction (p. 1) discusses fake news during the 2016 U.S. presidential election, but does not mention the Internet Research Agency or Russian efforts to influence the election. Searches for the agency, troll farms, and inflammatory online content found no passage addressing the attributed claim. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

May combine two claims.

Fake news and fabricated content propagate exponentially at the early stage of their creation and can cause significant loss in a short amount of time, including manipulating the outcome of political events.

“Similarly, Qian et al. (2018) assert that fake news and fabricated content propagate exponentially at the early stage of its creation and can cause a significant loss in a short amount of time (Friggeri et al. 2014) including manipulating the outcome of political events (Liu and Wu 2018; Bessi and Ferrara 2016).”

• cites Qian et al. 2018 sourceSupported
• cites Friggeri et al. 2014 sourcePartly supported

  The Introduction (page unavailable) says social networks can spread both true and false information quickly, and the Conclusion says dubious rumors can propagate rapidly. The paper does not establish exponential growth at the early stage or significant losses caused within a short time. The political-outcomes clause is attached to other citations, not Friggeri et al. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites Liu and Wu 2018 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

A malicious bot designed to make friends with as many accounts as possible requires a different detection approach than a bot designed to repeatedly post links to malicious websites.

“They note that a malicious bot designed to make friends with as many accounts as possible will require a different detection approach than a bot designed to repeatedly post links to malicious websites.”

• cites Ismailov et al. 2020 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Social bots amplify interactions with content as soon as it is created to make it look legitimate and facilitate its spread across social networks.

“First, they amplify interactions with content as soon as it is created to make it look legitimate and to facilitate its spread across social networks.”

• cites Shu et al. 2020a sourcePartly supported

  Section 3.3, “Bots on Social Media,” says bots amplify interactions as soon as content is created to help it go viral. It attributes increased perceived credibility to a different strategy: getting influential users to repost the content. The source therefore supports the timing and spread, but not the claim that the initial amplification itself makes the content look legitimate. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Feature extraction is regarded as a leading method for shrinking data volume in automated fake news detection.

“Feature extraction is considered as one of the best techniques to reduce data size in automatic fake news detection.”

• cites Yazdi et al. 2020 sourcePartly supported

  The introductory discussion says reducing dataset dimensions can lower computational complexity and improve fake-news classification. Section III describes feature selection—removing redundant features—not feature extraction. The copy also does not rank it as one of the best techniques. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Other researchers categorize false information on the web and social media based on intent and knowledge.

“Other researchers categorize false information on the web and social media based on its intent and knowledge (i.e., when there is a single ground truth).”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

Researchers have approached the fake news problem from many different perspectives and topics.

“The fake news problem has been addressed by researchers from various perspectives related to different topics.”

Classified as probably not a substantial claim.

The most used terms for referring to, categorizing and classifying fake news can be summarized using two key features: intent and authenticity of the news content.

“The terms most used in the literature to refer, categorize and classify fake news can be summarized and defined as shown in Table 3, in which we capture the similarities and show the differences between the different terms based on two common key features, which are the intent and the authenticity of the news content.”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Existing definitions agree that fake news is false or non-factual, but they disagree about whether related concepts such as satire, rumors, conspiracy theories, misinformation and hoaxes should be included.

“Then, other definitions were provided in the literature, but they all agree on the authenticity of fake news to be false (i.e., being non-factual). However, they disagree on the inclusion and exclusion of some related concepts such as satire, rumors, conspiracy theories, misinformation and hoaxes from the given definition.”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

There is still no agreed definition of the term fake news.

“Hence, there is still no agreed definition of the term “fake news.””

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

Various categorizations of fake news have been provided in the literature.

“Various categorizations of fake news have been provided in the literature.”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

Many terms and concepts in the literature refer to fake news.

“Moreover, we can find many terms and concepts in the literature that refer to fake news (Van der Linden et al. 2020; Molina et al. 2021)”

• cites Van der Linden et al. 2020 sourceNot checked — no full text of the source
• cites Molina et al. 2021 sourceNot checked — no full text of the source

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The literature agrees on the definitions of misinformation, disinformation and malinformation, but there is no agreed definition of information disorder, false information and fake news.

“The literature agrees on the definitions of the latter group, but there is still no agreed-upon definition of the first group.”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

The features used in the literature to define fake news include intent, authenticity and knowledge.

“In Fig. 3, we identify the different features used in the literature to define fake news (i.e., intent, authenticity and knowledge).”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Some definitions rely on both authenticity and intent, while others rely on either authenticity or intent.

“Hence, some definitions are based on two key features, which are authenticity and intent (i.e., news articles that are intentionally and verifiably false and could mislead readers). However, other definitions are based on either authenticity or intent.”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

Existing terms can be divided into general terms (information disorder, false information, fake news) and elementary terms (misinformation, disinformation, malinformation).

“The existing terms can be separated into two groups. The first group represents the general terms, which are information disorder, false information and fake news, each of which includes a subset of terms from the second group. The second group represents the elementary terms, which are misinformation, disinformation and malinformation.”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Some terms are explicitly used to refer to fake news (disinformation, misinformation and false information), while malinformation is not.

“Some of these terms are explicitly used to refer to fake news (i.e., disinformation, misinformation and false information), while others are not (i.e., malinformation).”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

Fake news can be divided into two major categories based on the studied perspective: intention or content.

“We can distinguish two major categories of fake news based on the studied perspective (i.e., intention or content) as shown in Fig. 4.”

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

May combine two claims.

Blockchain-based fake news solutions remain scarce and largely conceptual.

“However, the proposed blockchain approaches are few in number and they are fundamental and theoretical approaches.”

• AssumptionRests on an assumption: The reviewed body of literature is representative of the state of the art in automatic fake news detection.

Probably not a substantial claim: may not pass S3 (load-bearing).

Several social media platforms have incorporated fact-checking functions.

“Indeed, fact-checking features are being adopted by multiple online social network platforms. For instance, Facebook34 started addressing false information through independent fact-checkers in 2017, followed by Google35 the same year. Two years later, Instagram36 followed suit.”

May combine two claims.

A majority of studies treat fake news detection as a classification task and apply AI methods.

“Most researchers consider fake news detection as a classification problem and use artificial intelligence techniques, as shown in Fig. 8.”

• AssumptionRests on an assumption: The reviewed body of literature is representative of the state of the art in automatic fake news detection.

Probably not a substantial claim: may not pass S3 (load-bearing).

Most solutions mix several AI methods instead of using a single technique.

“Most of them combine many AI techniques in their solutions rather than relying on one specific approach.”

• AssumptionRests on an assumption: The reviewed body of literature is representative of the state of the art in automatic fake news detection.

May combine two claims.

Most article-content methods frame detection as classification and employ AI, including classical ML and deep learning.

“Most of the news content-based approaches consider fake news detection as a classification problem and they use AI techniques such as classical machine learning (e.g., regression, Bayesian) as well as deep learning (i.e., neural methods such as CNN and RNN).”

• AssumptionRests on an assumption: The reviewed body of literature is representative of the state of the art in automatic fake news detection.

Probably not a substantial claim: may not pass S2 (checkable).

Detecting fake news on online social networks has distinctive features and difficulties that make a solution far from simple.

“Furthermore, fake news detection on OSNs presents unique characteristics and challenges that make finding a solution anything but trivial.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Falsehood spreads significantly farther, faster, deeper and more broadly than truth across all categories of information.

“falsehood spreads significantly farther, faster, deeper and more broadly than the truth in all categories of information”

• cites Vosoughi et al. 2018 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Fake news, misinformation and disinformation belong to the broader landscape of online deception.

“fake news, misinformation and disinformation are indeed parts of the larger landscape of online deception (Hage et al. 2021).”

• cites Hage et al. 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Malicious attackers are using increasingly complex tools and strategies to deceive users.

“Hence, malicious attackers are using more complex tools and strategies to deceive users.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Online deception involves several factors: the deceiver, the purpose of the deception, the social media service, the deception technique, and the potential target.

“In the context of online deception, several factors need to be considered: the deceiver, the purpose or aim of the deception, the social media service, the deception technique and the potential target (Aïmeur et al. 2018; Hage et al. 2021).”

• cites Aïmeur et al. 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Hage et al. 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S4 (based).

Machine learning and deep learning techniques are used to deceive people by producing and spreading fake content.

“AI techniques such as machine learning and deep learning are leveraged to deceive people by creating and disseminating fake content.”

Probably not a substantial claim: may not pass S3 (load-bearing).

False news travels online six times faster than truthful content.

“the spread of false news online is six times faster than truthful content”

• cites Vosoughi et al. 2018 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Deception depends on people's trust, doubt and strong emotions, which can stop them from thinking and acting clearly.

“Deception relies on peoples’ trust, doubt and strong emotions that may prevent them from thinking and acting clearly (Aïmeur et al. 2018).”

• cites Aïmeur et al. 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Identifying fake news remains a difficult and unresolved problem.

“Fake news identification is still a complex unresolved issue.”

Probably not a substantial claim: may not pass S4 (based).

Fake content is crafted to look very much like the truth.

“the content is designed in a way to closely resemble the truth”

Probably not a substantial claim: may not pass S3 (load-bearing).

Seventy percent of users could not tell real news from fake news.

“70% of the users could not distinguish real from fake news (Vosoughi et al. 2018)”

• cites Vosoughi et al. 2018 sourceNot checked — no full text of the source

May combine two claims.

Fake news detection on social media is still at an early stage, and many difficult issues need deeper study.

“fake news detection on social media is still in the early age of development, and many challenging issues remain that require deeper investigation.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Deception leads people to do things they would not otherwise do.

“In other words, deception gets people to do things they would not otherwise do.”

May combine two claims.

Contrary to conventional wisdom, robots have accelerated the spread of real and fake news at the same rate, implying that fake news spreads more than the truth because humans, not robots, are more likely to spread it.

“Furthermore, Vosoughi et al. (Vosoughi et al. 2018) found that contrary to conventional wisdom, robots have accelerated the spread of real and fake news at the same rate, implying that fake news spreads more than the truth because humans, not robots, are more likely to spread it.”

• cites Vosoughi et al. 2018 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

People may choose misinformation because they engage in directionally motivated reasoning.

“However, it is also possible that people choose misinformation because they engage in directionally motivated reasoning (Badawy et al. 2019; Flynn et al. 2017).”

• cites Badawy et al. 2019 sourceNot checked — no full text of the source
• cites Flynn et al. 2017 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The purpose of clickbait is to increase traffic to a website.

“However, the goal behind using clickbait is to increase the traffic to a website (Zannettou et al. 2019).”

• cites Zannettou et al. 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Propaganda is a consequential type of false information because it can change the course of human history, for example by changing an election outcome.

“It is a consequential type of false information as it can change the course of human history (e.g., by changing the outcome of an election) (Zannettou et al. 2019).”

• cites Zannettou et al. 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Propaganda types of fake news can be debunked with manual fact-based detection models such as expert-based fact-checkers.

“Propaganda types of fake news can be debunked with manual fact-based detection models such as the use of expert-based fact-checkers (Collins et al. 2020).”

• cites Collins et al. 2020 sourceNot checked — no full text of the source

May combine two claims.

Overlap among fake news types has been proven, so false information can fall within multiple categories.

“Moreover, it is important to highlight that the existence of the overlap in the types of fake news mentioned above has been proven, thus it is possible to observe false information that may fall within multiple categories (Zannettou et al. 2019).”

• cites Zannettou et al. 2019 sourceNot checked — no full text of the source

May combine two claims.

Most audiences believed information presented in satirical form because comedians usually project mainstream news and frame it to suit their program.

“However, most audiences believed the information passed in this satirical form because the comedian usually projects news from mainstream media and frames them to suit their program (Collins et al. 2020).”

• cites Collins et al. 2020 sourceNot checked — no full text of the source

May combine two claims.

Online clients are normally vulnerable and generally perceive web-based social media as reliable.

“Online clients are normally vulnerable and will, in general, perceive web-based networking media as reliable, as reported by Abdullah-All-Tanvir et al. (2019), who propose to mechanize fake news recognition.”

• cites Abdullah-All-Tanvir et al. 2019 sourceNot checked — no full text of the source

May combine two claims.

Many people believe almost everything on the Internet, and those new to digital technology or with less expertise may be easily fooled.

“This is a complex issue because many people believe almost everything on the Internet and the ones who are new to digital technology or have less expertise may be easily fooled (Edgerly et al. 2020).”

• cites Edgerly et al. 2020 sourceNot checked — no full text of the source

May combine two claims.

People are often motivated to support and accept information that aligns with their preexisting viewpoints and beliefs, and to reject information that does not fit.

“Moreover, it has been widely proven (Metzger et al. 2020; Edgerly et al. 2020) that people are often motivated to support and accept information that goes with their preexisting viewpoints and beliefs, and reject information that does not fit in as well.”

• cites Metzger et al. 2020 sourceNot checked — no full text of the source
• cites Edgerly et al. 2020 sourceNot checked — no full text of the source

May combine two claims.

Headlines associated with a high number of likes increased credibility, enhancing detection accuracy for real news but undermining accuracy for fake news.

“Results showed that headlines associated with a high number of likes increased credibility, thereby enhancing detection accuracy for real news but undermining accuracy for fake news.”

• cites Luo et al. 2022 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Zhang and Ghorbani (2020) proposed a classification that combines content and intent and distinguishes physical from non-physical news content.

“Recently, another classification was proposed by Zhang and Ghorbani (2020). It is based on the combination of content and intent to categorize fake news. They distinguish physical news content and non-physical news content from fake news.”

• cites Zhang and Ghorbani (2020) sourceNot checked — no full text of the source

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

Probably not a substantial claim: may not pass S3 (load-bearing).

Falling for disinformation may be explained by a lack of critical thinking and of the need for evidence supporting information.

“Falling for disinformation may also be explained by a lack of critical thinking and of the need for evidence that supports information (Vilmer et al. 2018; Badawy et al. 2019).”

• cites Vilmer et al. 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Badawy et al. 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Fake news tends to have more complicated stories and hardly ever makes references.

“Abdullah-All-Tanvir et al. (2020) reported that fake news tends to have more complicated stories and hardly ever make any references.”

• cites Abdullah-All-Tanvir et al. 2020 sourceNot checked — no full text of the source

May combine two claims.

Misinformed people are a greater problem than uninformed people because the former hold inaccurate opinions that are harder to correct.

“Moreover, misinformed people are a greater problem than uninformed people (Kuklinski et al. 2000), because the former hold inaccurate opinions (which may concern politics, climate change, medicine) that are harder to correct.”

• cites Kuklinski et al. 2000 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Participants exhibit higher intent to verify news only when they believe the headline to be true, which is predicted by perceived congruence with preexisting ideological tendencies.

“They concluded that participants exhibit higher intent to verify the news only when they believe the headline to be true, which is predicted by perceived congruence with preexisting ideological tendencies.”

• cites Edgerly et al. 2020 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Online astroturfing can be divided into political astroturfing, corporate astroturfing, and astroturfing in e-commerce or online services.

“Astroturfing can affect different domains of interest, based on which online astroturfing can be mainly divided into political astroturfing, corporate astroturfing and astroturfing in e-commerce or online services (Mahbub et al. 2019).”

• cites Mahbub et al. 2019 sourceNot checked — no full text of the source

May combine two claims.

Different forms of fake news content have differing effects on credibility judgments and sharing intentions, and this variation affects how fake news spreads on online social networks.

“The effects of these forms of fake news content vary on the credibility assessment, as well as sharing intentions which influences the spread of fake news on OSNs.”

May combine two claims.

Conspiracy theories are widespread in contemporary democracies and have major consequences.

“Conspiracy theories are widespread in contemporary democracies (Sutton and Douglas 2020), and they have major consequences.”

• cites Sutton and Douglas 2020 sourceNot checked — no full text of the source

May combine two claims.

Detecting false information is not straightforward because fake news appears in multiple types and forms, making a single global solution hard to define.

“Moreover, fake news appears in multiple types and forms, which makes it hard and challenging to define a single global solution able to capture and deal with the disseminated content.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Humans are naturally not very good at differentiating real information from fake information.

“Consequently, they deduce that humans are naturally not very good at differentiating real information from fake information.”

May combine two claims.

Deepfake videos and GAN-generated fake images are AI-based machine-generated fake content that unsophisticated social network users find hard to identify.

“we can mention as examples deepfake videos (Yang et al. 2019b) and GAN-generated fake images (Zhang et al. 2019b), which are artificial intelligence-based machine-generated fake content that are hard for unsophisticated social network users to identify.”

• cites Yang et al. 2019bNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Zhang et al. 2019bNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• AssumptionRests on an assumption: Despite the absence of an agreed definition, fake news is a coherent enough category to review and classify.

May combine two claims.

In addition to bots causing most misrepresentations, specific individuals also contribute a large share of the issue.

“It is worth noting that in addition to bots causing the outpouring of the majority of the misrepresentations, specific individuals are also contributing a large share of this issue (Abdullah-All-Tanvir et al. 2019).”

• cites Abdullah-All-Tanvir et al. 2019 sourceNot checked — no full text of the source

May combine two claims.

Social bots tend to connect with legitimate users instead of other bots, act like humans with fewer words and fewer followers, and this contributes to forwarding fake news.

“Social bots tend to connect with legitimate users instead of other bots. They try to act like a human with fewer words and fewer followers on social media. This contributes to the forwarding of fake news (Jiang et al. 2019).”

• cites Jiang et al. 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Social bots played a disproportionate role in spreading articles from low-credibility sources during and after the 2016 US presidential campaign by amplifying such content early and targeting users with many followers through replies and mentions to expose them and induce sharing.

“They found that social bots played a disproportionate role in spreading articles from low-credibility sources by amplifying such content in the early spreading moments and targeting users with many followers through replies and mentions to expose them to this content and induce them to share it.”

• cites Shao et al. 2018aNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Fake news detection solutions can be categorized as news content-based approaches, social context-based approaches divided into network and user-based approaches, and hybrid approaches that combine content-based and contextual approaches.

“fake news detection solutions can be categorized as news content-based approaches, the social context-based approaches that can be divided into network and user-based approaches, and hybrid approaches.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Review bombing refers to coordinated groups of people massively performing the same negative actions online on a video, game, post, product, etc., to reduce its aggregate review score.

“Review bombing refers to coordinated groups of people massively performing the same negative actions online (e.g., dislike, negative review/comment) on an online video, game, post, product, etc., in order to reduce its aggregate review score.”

• cites Moro and Birt 2022 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

There is a difference between bot-generated and human-written clickbait.

“Moreover, there is a difference between bot-generated and human-written clickbait (Le et al. 2019).”

• cites Le et al. 2019Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

The dynamic nature of online social platforms results in continued rapid and exponential propagation of fake content and remains a major challenge requiring further investigation.

“Therefore, it is crucial to note that the dynamic nature of the various online social platforms, which results in the continued rapid and exponential propagation of such fake content, remains a major challenge that requires further investigation while defining innovative solutions for fake news detection.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Twitter suspended accounts connected to the IRA and deleted 200,000 tweets from Russian trolls.

“As a result, Twitter suspended accounts connected to the IRA and deleted 200,000 tweets from Russian trolls (Jamieson 2020).”

• cites Jamieson 2020 sourceNot checked — no full text of the source

May combine two claims.

People alone cannot handle fake news identification, since it is costly and slow and cannot stop rapid dissemination.

“However, fake news identification cannot be addressed solely through human force since it demands a lot of effort in terms of time and cost, and it is ineffective in terms of preventing the fast spread of fake content.”

Classified as probably not a substantial claim.

It is almost impossible to manually detect the sources and authenticity of fake news effectively and efficiently due to its fast circulation in a small amount of time.

“Lastly, Abdullah-All-Tanvir et al. (2020) note that it is almost impossible to manually detect the sources and authenticity of fake news effectively and efficiently, due to its fast circulation in such a small amount of time.”

• cites Abdullah-All-Tanvir et al. 2020 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

A recent classification by Sharma et al. uses three categories of fake news identification methods, each further divided based on the type of existing methods: content-based, feedback-based, and intervention-based.

“A recent classification by Sharma et al. (2019) uses three categories of fake news identification methods. Each category is further divided based on the type of existing methods (i.e., content-based, feedback-based and intervention-based methods).”

• cites Sharma et al. 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Hierarchical propagation network features from structural, temporal, and linguistic perspectives have potential and effectiveness for fake news detection.

“They demonstrated the potential of using these features to detect fake news and they showed their effectiveness for fake news detection as well.”

• cites Shu et al. 2020bNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Deep learning can derive hidden representations from basic inputs across different contexts and content, unlike traditional ML that uses handcrafted features.

“Unlike traditional machine learning approaches, which are based on manually crafted features, deep learning approaches can learn hidden representations from simpler inputs both in context and content variations (Bondielli and Marcelloni 2019).”

• cites Bondielli and Marcelloni 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Fake news content is produced and disseminated faster and easier through social media because of low barriers that prevent doing so.

“found that such content is produced and disseminated faster and easier through social media because of the low barriers that prevent doing so.”

• cites Shu et al. 2020cNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

Journalists usually carry out fact-checking by hand to test whether a statement is true.

“The fact-checking task is commonly manually performed by journalists to verify the truthfulness of a given claim.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Few research studies have focused on automatic detection of fake news using artificial intelligence techniques.

“Few of them have focused on the automatic detection of fake news using artificial intelligence techniques.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Review bombers can be both humans and bots coordinated to cause harm and mislead people by falsifying facts.

“The review bombers can be both humans and bots coordinated in order to cause harm and mislead people by falsifying facts.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Bot farms or troll farms are groups of organized individuals engaging in trolling or bot-like promotion of narratives in a coordinated fashion, hired to massively spread fake news or harmful content.

“It is not social bots, but it is a group of organized individuals engaging in trolling or bot-like promotion of narratives in a coordinated fashion (Wardle 2018) hired to massively spread fake news or any other harmful content.”

• cites Wardle 2018 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based).

Tracking past user conduct, such as expressed feelings or views on news, may aid early detection and reduction of fake news.

“Moreover, capturing users’ historical behavior, including their emotions and/or opinions toward news content, can help in the early detection and mitigation of fake news.”

May combine two claims.

Methods using context examine data around the article and may work well where text-based content methods struggle.

“The context-based approaches explore the surrounding data outside of the news content, which can be an effective direction and has some advantages in areas where the content approaches based on text classification can run into issues.”

Classified as probably not a substantial claim.

Data about users and their networks can supply helpful extra signals that improve detection accuracy.

“However, contextual information derived from social media users and from the network can provide useful auxiliary information to increase detection accuracy.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The selection of information sources and search terms could have biased this review.

“Our choice of sources of information and search criteria may have introduced biases in our research. If so, it would be desirable to identify those biases and mitigate them.”

Classified as probably not a substantial claim.

Recording user traits and how they act on shared posts may be central to detecting fake news.

“Specifically, capturing users’ characteristics and users’ behavior toward shared content can be a key task for fake news detection.”

Classified as probably not a substantial claim.

Article text is the primary place to look for signals that separate false from true reporting.

“News content is the fundamental source to find clues to distinguish fake from real content.”

Classified as probably not a substantial claim.

Adversarial methods like GAN and SeqGAN may ease dataset shortages by generating synthetic data for training stronger detectors.

“Furthermore, adversarial learning techniques (e.g., GAN, SeqGAN) can be considered as a promising direction for mitigating the lack and scarcity of available datasets by providing machine-generated data that can be used to train and build robust systems to detect the fake examples from the real ones.”

May combine two claims.

Contextual studies rely heavily on advanced ML for feature extraction and neglect web search and crowdsourcing, which could speed up early detection.

“Moreover, from a technical perspective, they are limited to the use of sophisticated machine learning techniques for feature extraction, and they ignore the usefulness of results coming from techniques such as web search and crowdsourcing which may save much time and help in the early detection and identification of fake content.”

May combine two claims.

Content-analysis methods are less affected by early detection, yet they still lack verification information while a story is just beginning to spread.

“Although approaches that detect fake news based on content analysis face this issue less, they are still limited by the lack of information required for verification when the news is in its early stage of spread.”