The COVID-19 social media infodemic

The COVID-19 social media infodemic Matteo Cinelli Walter Quattrociocchi Alessandro Galeazzi Carlo Michele Valensise Emanuele Brugnoli Ana Lucia Schmidt Paola Zola Fabiana Zollo Antonio Scala

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

The COVID-19 social media infodemic 2020 Scientific Reports 10.1038/s41598-020-73510-5 cinelli2020-the-covid-19-social-media-infodemi cinelli2020-the-covid-19-social-media-infodemi.txt, the project's staged copy 4245c9f5a259e994cf1601601d0bd9840460e0fd cc-by social-psychology

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them how COVID-19 information spread on Twitter, Instagram, YouTube, Reddit and Gab over the same 45 days, and whether content from questionable outlets travels differently from content from reliable ones Quotation withheld because this record is not cleared for redistribution. whether the diffusion of misinformation is a property of the content or of the platform it is on Quotation withheld because this record is not cleared for redistribution. a five-platform comparison over one keyword set, with an epidemic model fitted per platform, showing that reliable and questionable content share a growth pattern whose ratio depends entirely on the platform Quotation withheld because this record is not cleared for redistribution. existence Quotation withheld because this record is not cleared for redistribution. an infodemic - and a quantity whose value is a property of the environment measured rather than of the thing measured Quotation withheld because this record is not cleared for redistribution. observational-study observational-study Quotation withheld because this record is not cleared for redistribution. posts and reactions from questionable outlets grow at the same pace as those from reliable ones inside a platform, so the questionable series is a re-scaling of the reliable one. THE ONLY THING THAT VARIES IS THE SCALING FACTOR, AND IT VARIES BY PLATFORM - which means a single-platform study measures its platform, not the phenomenon the same curve, a different multiplier Quotation withheld because this record is not cleared for redistribution. Gab posts about 70% as much questionable as reliable content and reacts to it about three times more. Counting posts and counting reactions therefore rank the platforms differently, and the amplification factor exists precisely to separate them volume and reaction come apart Quotation withheld because this record is not cleared for redistribution. reliability comes from Media Bias/Fact Check at the level of the news outlet. Every article from a listed outlet inherits its label, and every article from an unlisted one is invisible. The paper states the limit and prints the six criteria rather than citing the list, which is what makes the boundary checkable the label is on the outlet, and the authors say so Quotation withheld because this record is not cleared for redistribution. every platform's data was gathered by the same Google-Trends-derived keyword list, and on Gab the hashtags were then manually browsed and selected by meaning. What is in the dataset is what those keywords caught, and the comparability across five platforms rests on the list being applied identically the keyword set is the aperture Quotation withheld because this record is not cleared for redistribution. R0 is the number of secondary cases an infectious individual creates. Here the infection is posting about COVID-19, so R0 > 1 means the topic keeps spreading. The metaphor imports a threshold at 1 that the social process did not have an epidemiological model borrowed for language Quotation withheld because this record is not cleared for redistribution. reliability is labelled at the level of the news outlet rather than the single article, and the authors name this as the limit of the approach Quotation withheld because this record is not cleared for redistribution. the definition of `fake news` may be `inadequate, since political debate often resorts to labelling opposite news as unreliable or fake` Quotation withheld because this record is not cleared for redistribution. Twitter data before 27 January could not be gathered because of API limitations, so that platform's series starts later than the others Quotation withheld because this record is not cleared for redistribution. Gab has no official API guide, so hashtags were manually browsed and selected by meaning rather than by a documented procedure Quotation withheld because this record is not cleared for redistribution. the whole dataset is bounded by a keyword list derived from Google Trends Quotation withheld because this record is not cleared for redistribution. the basic reproduction number from epidemiology, applied to users who begin posting about a topic rather than to infected individuals Quotation withheld because this record is not cleared for redistribution. a third-party fact-checking organisation's outlet ratings, used as the ground truth for content reliability Quotation withheld because this record is not cleared for redistribution. five platforms analysed over the same time window with the same keyword set, rather than one eliminates: a platform-specific effect being reported as a property of misinformation - which the paper names as the flaw in single-platform studies · outcome: held Quotation withheld because this record is not cleared for redistribution. the growth of reliable-source content plotted against questionable-source content within each platform, rather than questionable content alone eliminates: a rise in misinformation that is really a rise in posting - the two grow at the same pace, so the ratio is the only thing that carries information · outcome: held Quotation withheld because this record is not cleared for redistribution. the amplification factor defined and measured separately from post volume eliminates: posting and reacting being conflated - on Gab they point in opposite directions · outcome: held Quotation withheld because this record is not cleared for redistribution. the six criteria for a questionable source printed in the paper rather than cited to the fact-checking organisation eliminates: a category readers cannot inspect - the boundary is stated even though the labels come from a third party · outcome: held Quotation withheld because this record is not cleared for redistribution. keyword-driven collection applied identically across platforms, with the Gab hashtags manually browsed because no documented API guide exists eliminates: a cross-platform difference that is really a difference in how each platform was searched · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. questionable questionable Quotation withheld because this record is not cleared for redistribution. reliable reliable Quotation withheld because this record is not cleared for redistribution. news outlets news outlets Quotation withheld because this record is not cleared for redistribution. single articles single articles Quotation withheld because this record is not cleared for redistribution. platform platform Quotation withheld because this record is not cleared for redistribution. Gab Gab Quotation withheld because this record is not cleared for redistribution. volume volume Quotation withheld because this record is not cleared for redistribution. reactions reactions Quotation withheld because this record is not cleared for redistribution. amplification factor amplification factor Quotation withheld because this record is not cleared for redistribution. fake news fake news Quotation withheld because this record is not cleared for redistribution. keywords keywords Quotation withheld because this record is not cleared for redistribution. post post Quotation withheld because this record is not cleared for redistribution. diffusion of information diffusion of information Quotation withheld because this record is not cleared for redistribution. massive data analysis massive data analysis Quotation withheld because this record is not cleared for redistribution. Twitter Twitter Quotation withheld because this record is not cleared for redistribution. Instagram Instagram Quotation withheld because this record is not cleared for redistribution. YouTube YouTube Quotation withheld because this record is not cleared for redistribution. Reddit Reddit Quotation withheld because this record is not cleared for redistribution. engagement engagement Quotation withheld because this record is not cleared for redistribution. interest interest Quotation withheld because this record is not cleared for redistribution. epidemic models epidemic models Quotation withheld because this record is not cleared for redistribution. basic reproduction number R0 basic reproduction number R0 Quotation withheld because this record is not cleared for redistribution. misinformation misinformation Quotation withheld because this record is not cleared for redistribution. spreading patterns spreading patterns Quotation withheld because this record is not cleared for redistribution. rumors’ amplification rumors’ amplification Quotation withheld because this record is not cleared for redistribution. SARS-CoV-2 virus outbreak SARS-CoV-2 virus outbreak Quotation withheld because this record is not cleared for redistribution. global threat global threat Quotation withheld because this record is not cleared for redistribution. disintermediated news cycle disintermediated news cycle Quotation withheld because this record is not cleared for redistribution. infodemic infodemic Quotation withheld because this record is not cleared for redistribution. misinformation phenomena misinformation phenomena Quotation withheld because this record is not cleared for redistribution. social response social response Quotation withheld because this record is not cleared for redistribution. information is consumed and reported information is consumed and reported Quotation withheld because this record is not cleared for redistribution. behavioral response of the population behavioral response of the population Quotation withheld because this record is not cleared for redistribution. public health interventions public health interventions Quotation withheld because this record is not cleared for redistribution. communication dynamics communication dynamics Quotation withheld because this record is not cleared for redistribution. users’ preferences and attitudes users’ preferences and attitudes Quotation withheld because this record is not cleared for redistribution. algorithms algorithms Quotation withheld because this record is not cleared for redistribution. content promotion content promotion Quotation withheld because this record is not cleared for redistribution. social perceptions social perceptions Quotation withheld because this record is not cleared for redistribution. framing of narratives framing of narratives Quotation withheld because this record is not cleared for redistribution. policy-making policy-making Quotation withheld because this record is not cleared for redistribution. polarized groups polarized groups Quotation withheld because this record is not cleared for redistribution. shared narratives shared narratives Quotation withheld because this record is not cleared for redistribution. comparative analysis comparative analysis Quotation withheld because this record is not cleared for redistribution. five social media platforms five social media platforms Quotation withheld because this record is not cleared for redistribution. 8 million comments and posts 8 million comments and posts Quotation withheld because this record is not cleared for redistribution. 45 days 45 days Quotation withheld because this record is not cleared for redistribution. secondary cases secondary cases Quotation withheld because this record is not cleared for redistribution. “infectious” individual “infectious” individual Quotation withheld because this record is not cleared for redistribution. R0 = 1 R0 = 1 Quotation withheld because this record is not cleared for redistribution. mainstream platforms mainstream platforms Quotation withheld because this record is not cleared for redistribution. less regulated social media platforms less regulated social media platforms Quotation withheld because this record is not cleared for redistribution. crowdfunded social media crowdfunded social media Quotation withheld because this record is not cleared for redistribution. social news aggregation social news aggregation Quotation withheld because this record is not cleared for redistribution. collective filtering of information collective filtering of information Quotation withheld because this record is not cleared for redistribution. Google Trends’ COVID-19 related queries Google Trends’ COVID-19 related queries Quotation withheld because this record is not cleared for redistribution. 1,342,103 posts 1,342,103 posts Quotation withheld because this record is not cleared for redistribution. 7,465,721 comments 7,465,721 comments Quotation withheld because this record is not cleared for redistribution. 3,734,815 users 3,734,815 users Quotation withheld because this record is not cleared for redistribution. interactions interactions Quotation withheld because this record is not cleared for redistribution. users’ activity users’ activity Quotation withheld because this record is not cleared for redistribution. long tail long tail Quotation withheld because this record is not cleared for redistribution. dynamics of reactions and content consumption dynamics of reactions and content consumption Quotation withheld because this record is not cleared for redistribution. Natural Language Processing techniques Natural Language Processing techniques Quotation withheld because this record is not cleared for redistribution. word embedding word embedding Quotation withheld because this record is not cleared for redistribution. word vector representation word vector representation Quotation withheld because this record is not cleared for redistribution. clustering procedures clustering procedures Quotation withheld because this record is not cleared for redistribution. cumulative number of new posts each day cumulative number of new posts each day Quotation withheld because this record is not cleared for redistribution. first situation report first situation report Quotation withheld because this record is not cleared for redistribution. specific timings for content consumption specific timings for content consumption Quotation withheld because this record is not cleared for redistribution. interaction mechanisms interaction mechanisms Quotation withheld because this record is not cleared for redistribution. standard epidemic models standard epidemic models Quotation withheld because this record is not cleared for redistribution. monotonic increasing trend monotonic increasing trend Quotation withheld because this record is not cleared for redistribution. growth mechanisms growth mechanisms Quotation withheld because this record is not cleared for redistribution. phenomenological models phenomenological models Quotation withheld because this record is not cleared for redistribution. basic reproduction number R0 basic reproduction number R0 Quotation withheld because this record is not cleared for redistribution. infective process infective process Quotation withheld because this record is not cleared for redistribution. EXP model EXP model Quotation withheld because this record is not cleared for redistribution. SIR (Susceptible, Infected, Recovered) compartmental model SIR (Susceptible, Infected, Recovered) compartmental model Quotation withheld because this record is not cleared for redistribution. supercritical supercritical Quotation withheld because this record is not cleared for redistribution. engagement rate engagement rate Quotation withheld because this record is not cleared for redistribution. social contagion phenomena social contagion phenomena Quotation withheld because this record is not cleared for redistribution. abrupt jump abrupt jump Quotation withheld because this record is not cleared for redistribution. continuous models continuous models Quotation withheld because this record is not cleared for redistribution. Media Bias/Fact Check Media Bias/Fact Check Quotation withheld because this record is not cleared for redistribution. extreme bias extreme bias Quotation withheld because this record is not cleared for redistribution. propaganda/conspiracies propaganda/conspiracies Quotation withheld because this record is not cleared for redistribution. poor or no sourcing poor or no sourcing Quotation withheld because this record is not cleared for redistribution. lack of transparency lack of transparency Quotation withheld because this record is not cleared for redistribution. overall reactions overall reactions Quotation withheld because this record is not cleared for redistribution. strong linear correlation strong linear correlation Quotation withheld because this record is not cleared for redistribution. growth dynamics growth dynamics Quotation withheld because this record is not cleared for redistribution. re-scaled version re-scaled version Quotation withheld because this record is not cleared for redistribution. re-scaling factor ρ re-scaling factor ρ Quotation withheld because this record is not cleared for redistribution. mainstream social media mainstream social media Quotation withheld because this record is not cleared for redistribution. coefficient of relative amplification coefficient of relative amplification Quotation withheld because this record is not cleared for redistribution. permeability of a platform permeability of a platform Quotation withheld because this record is not cleared for redistribution. Twitter is the most neutral Twitter is the most neutral Quotation withheld because this record is not cleared for redistribution. YouTube amplifies questionable sources less YouTube amplifies questionable sources less Quotation withheld because this record is not cleared for redistribution. Reddit reduces the impact of questionable sources Reddit reduces the impact of questionable sources Quotation withheld because this record is not cleared for redistribution. Gab strongly amplifies them Gab strongly amplifies them Quotation withheld because this record is not cleared for redistribution. interaction paradigm interaction paradigm Quotation withheld because this record is not cleared for redistribution. interaction patterns of groups of users interaction patterns of groups of users Quotation withheld because this record is not cleared for redistribution. tailored communication strategies tailored communication strategies Quotation withheld because this record is not cleared for redistribution. Reddit dataset Reddit dataset Quotation withheld because this record is not cleared for redistribution. YouTube Data API YouTube Data API Quotation withheld because this record is not cleared for redistribution. Twitter API Twitter API Quotation withheld because this record is not cleared for redistribution. stream API stream API Quotation withheld because this record is not cleared for redistribution. search API search API Quotation withheld because this record is not cleared for redistribution. Instagram Dataset Instagram Dataset Quotation withheld because this record is not cleared for redistribution. URLs URLs Quotation withheld because this record is not cleared for redistribution. MediaBias/FactCheck (MBFC) MediaBias/FactCheck (MBFC) Quotation withheld because this record is not cleared for redistribution. Factual/Sourcing Factual/Sourcing Quotation withheld because this record is not cleared for redistribution. Conspiracy-Pseudoscience Conspiracy-Pseudoscience Quotation withheld because this record is not cleared for redistribution. Pro-Science Pro-Science Quotation withheld because this record is not cleared for redistribution. binary label binary label Quotation withheld because this record is not cleared for redistribution. matching ability matching ability Quotation withheld because this record is not cleared for redistribution. Partitioning Around Medoids (PAM) algorithm Partitioning Around Medoids (PAM) algorithm Quotation withheld because this record is not cleared for redistribution. Skip-gram model Skip-gram model Quotation withheld because this record is not cleared for redistribution. distributed representations of words distributed representations of words Quotation withheld because this record is not cleared for redistribution. Latent Semantic Analysis (LSA) Latent Semantic Analysis (LSA) Quotation withheld because this record is not cleared for redistribution. Latent Dirichlet Allocation (LDA) Latent Dirichlet Allocation (LDA) Quotation withheld because this record is not cleared for redistribution. linear regularities among words linear regularities among words Quotation withheld because this record is not cleared for redistribution. English English Quotation withheld because this record is not cleared for redistribution. HTML code, URLs and email addresses HTML code, URLs and email addresses Quotation withheld because this record is not cleared for redistribution. stop-words stop-words Quotation withheld because this record is not cleared for redistribution. cosine distance matrix cosine distance matrix Quotation withheld because this record is not cleared for redistribution. average silhouette width average silhouette width Quotation withheld because this record is not cleared for redistribution. average pairwise Jaccard similarity average pairwise Jaccard similarity Quotation withheld because this record is not cleared for redistribution. word clouds word clouds Quotation withheld because this record is not cleared for redistribution. mechanistic models mechanistic models Quotation withheld because this record is not cleared for redistribution. adjusted exponential model adjusted exponential model Quotation withheld because this record is not cleared for redistribution. data-scarce settings data-scarce settings Quotation withheld because this record is not cleared for redistribution. classical SIR model classical SIR model Quotation withheld because this record is not cleared for redistribution. susceptible population susceptible population Quotation withheld because this record is not cleared for redistribution. infected individuals infected individuals Quotation withheld because this record is not cleared for redistribution. rate β rate β Quotation withheld because this record is not cleared for redistribution. γ γ Quotation withheld because this record is not cleared for redistribution. linear regression coefficients linear regression coefficients Quotation withheld because this record is not cleared for redistribution. R2 values R2 values Quotation withheld because this record is not cleared for redistribution. average number of secondary cases average number of secondary cases Quotation withheld because this record is not cleared for redistribution. Word embeddings Word embeddings Quotation withheld because this record is not cleared for redistribution. EXP EXP Quotation withheld because this record is not cleared for redistribution. E U /E R E U /E R Quotation withheld because this record is not cleared for redistribution. interaction paradigm imposed by the specific social media interaction paradigm imposed by the specific social media Quotation withheld because this record is not cleared for redistribution. information spreading information spreading Quotation withheld because this record is not cleared for redistribution. perils of misinformation phenomena perils of misinformation phenomena Quotation withheld because this record is not cleared for redistribution. dataset dataset Quotation withheld because this record is not cleared for redistribution. R0 > 1 R0 > 1 Quotation withheld because this record is not cleared for redistribution. R0 R0 Quotation withheld because this record is not cleared for redistribution. audience and interaction mechanisms audience and interaction mechanisms Quotation withheld because this record is not cleared for redistribution. Questionable Questionable Quotation withheld because this record is not cleared for redistribution. cluster words cluster words Quotation withheld because this record is not cleared for redistribution. linear relationship between reliable and questionable sources linear relationship between reliable and questionable sources Quotation withheld because this record is not cleared for redistribution. adjusted exponential model → EXP The paper indicates the adjusted exponential model as EXP; the gate's same-as cues do not include this wording, so the alias remains unhoused. Quotation withheld because this record is not cleared for redistribution. algorithms → content promotion Algorithms mediate content promotion; mediates is absent from the eight governed types. Quotation withheld because this record is not cleared for redistribution. Google Trends’ COVID-19 related queries → dataset The query set filters which platform content enters the dataset; selects-or-filters-into is absent from the governed types. Quotation withheld because this record is not cleared for redistribution. R0 > 1 → infodemic R0 above one indicates the possibility of an infodemic; indicator-of is absent from the eight governed types. Quotation withheld because this record is not cleared for redistribution. R0 → engagement rate R0 is used as a proxy for engagement rate; proxy-for is absent from the eight governed types. Quotation withheld because this record is not cleared for redistribution. specific timings for content consumption → audience and interaction mechanisms Platform timing patterns may depend on audience and social/algorithmic interaction mechanisms; depends-on is absent and the statement is hedged. Quotation withheld because this record is not cleared for redistribution. growth dynamics → re-scaling factor ρ Questionable-source growth is a rescaled version of reliable-source growth, with rho providing the platform-dependent scale; scaled-by is absent. Quotation withheld because this record is not cleared for redistribution. Conspiracy-Pseudoscience → Questionable MBFC outlets in Conspiracy-Pseudoscience are assigned the binary Questionable label; labelled-as by a classifier is absent from the governed types. Quotation withheld because this record is not cleared for redistribution. Partitioning Around Medoids (PAM) algorithm → cluster words PAM is used to cluster words from their embedding vectors; used-for is absent from the governed types. Quotation withheld because this record is not cleared for redistribution. R2 values → linear relationship between reliable and questionable sources High R-squared values confirm the linear source-class diffusion relationship; supports-or-confirms is absent from the eight types. Quotation withheld because this record is not cleared for redistribution. a questionable source - AS DEFINED HERE: a news OUTLET systematically showing extreme bias, promotion of propaganda or conspiracies, poor sourcing, unverifiable information, a lack of transparency, or fake news Quotation withheld because this record is not cleared for redistribution. a questionable source - WHAT THE LABEL CANNOT SEE: the individual article. The authors name labelling outlets rather than single articles as the limit of the approach, in the sentence that adopts it Quotation withheld because this record is not cleared for redistribution. `fake news` - AS A TERM: possibly `inadequate, since political debate often resorts to labelling opposite news as unreliable or fake` Quotation withheld because this record is not cleared for redistribution. the spread of misinformation - AS A PATTERN: the same growth dynamics as reliable content, re-scaled Quotation withheld because this record is not cleared for redistribution. the spread of misinformation - AS A VOLUME: a ratio strongly dependent on the platform, with Gab's reactions running three times the reliable volume Quotation withheld because this record is not cleared for redistribution. R0 - AS AN EPIDEMIC PARAMETER: the average number of secondary cases one infectious individual creates, with 1 as the threshold Quotation withheld because this record is not cleared for redistribution. Infodemic - misinformation during outbreak management can fragment social response and interfere with disease countermeasures Quotation withheld because this record is not cleared for redistribution. Cross-platform unit of comparison - the same COVID-19 argument is observed in environments with different audiences and interaction settings Quotation withheld because this record is not cleared for redistribution. Engagement distribution - all five platforms show long-tailed user activity even though their interaction volumes differ Quotation withheld because this record is not cleared for redistribution. Epidemic analogy - exposure precedes first posting, an already-posting user is infectious, and R0 above one denotes possible infodemic growth Quotation withheld because this record is not cleared for redistribution. Model limitation - Instagram's abrupt growth jump drives an implausibly large SIR R0 and cannot be represented by a continuous epidemic model Quotation withheld because this record is not cleared for redistribution. Source-class growth - questionable and reliable content share a linear growth pace within a platform but differ by a platform-specific rescaling factor Quotation withheld because this record is not cleared for redistribution. Volume versus amplification - Gab has lower questionable posting volume than reliable volume while questionable reactions are several times larger Quotation withheld because this record is not cleared for redistribution. Relative amplification - alpha above one favors questionable sources, below one favors reliable sources, and near one is neutral Quotation withheld because this record is not cleared for redistribution. Outlet-label aperture - only posts with external URLs matching the MBFC list can receive the study's Questionable or Reliable classification Quotation withheld because this record is not cleared for redistribution. Phenomenological versus mechanistic modeling - one reproduces empirical curves without a growth mechanism while the other tries to encode the mechanism Quotation withheld because this record is not cleared for redistribution. critiques → single-platform studies of misinformation diffusion, on the grounds that the dynamics may be particular to the environment Quotation withheld because this record is not cleared for redistribution. qualifies → the finding that fake news spreads faster and wider than fact-based news, by noting it might be a platform-specific effect Quotation withheld because this record is not cleared for redistribution. qualifies → its own reliability labels, by naming outlet-level rather than article-level tagging as the limit and printing the criteria Quotation withheld because this record is not cleared for redistribution. extends → epidemiological modelling, by fitting a basic reproduction number to the spread of a topic rather than of a disease Quotation withheld because this record is not cleared for redistribution. news outlets —distinct-from→ single articles news outlets single articles differs in: what unit the reliability judgement is attached to · the relation itself is separately evidenced · strength of the relation: 0.95 · differs in (stated by the tagger, no quotation): what unit the reliability judgement is attached to · the relation stated: reliability is tagged at the outlet level from a third-party fact-checker; other studies label individual articles, and the authors name the difference as the limit of their approach Quotation withheld because this record is not cleared for redistribution. volume —distinct-from→ reactions volume reactions differs in: whether the platform's difference is in what is posted or in what is reacted to · the relation itself is separately evidenced · strength of the relation: 0.95 · differs in (stated by the tagger, no quotation): whether the platform's difference is in what is posted or in what is reacted to · the relation stated: on Gab the volume of questionable posts is about 70% of reliable ones while their reactions are about three times bigger - the volume and the response come apart, and only on that platform Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span compares post volume with reactions; distinct-from is loose. Quotation withheld because this record is not cleared for redistribution. amplification factor —operationalised-as→ post amplification factor post the relation itself is separately evidenced · strength of the relation: 0.95 · the relation stated: amplification is defined as the average number of reactions a post receives, which separates how much is published from how much is responded to Audit 2026-09-28: DOUBTFUL - read against its span and kept, but E is the average number of reactions to a post; the endpoint 'post' drops 'reactions'. Quotation withheld because this record is not cleared for redistribution. Gab —is-a→ crowdfunded social media Gab crowdfunded social media the relation itself is separately evidenced · strength of the relation: 1 · note: The paper directly classifies Gab as a crowdfunded social medium. Quotation withheld because this record is not cleared for redistribution. Reddit —is-a→ social news aggregation Reddit social news aggregation the relation itself is separately evidenced · strength of the relation: 1 · note: The paper directly classifies Reddit as social news aggregation. Quotation withheld because this record is not cleared for redistribution. R0 —operationalised-as→ average number of secondary cases R0 average number of secondary cases the relation itself is separately evidenced · strength of the relation: 1 · note: The i.e. clause supplies the study's platform-level operational definition. Quotation withheld because this record is not cleared for redistribution. interactions —operationalised-as→ engagement interactions engagement the relation itself is separately evidenced · strength of the relation: 1 · note: The study equates its interaction measure with engagement using i.e. Quotation withheld because this record is not cleared for redistribution. word embedding —operationalised-as→ word vector representation word embedding word vector representation the relation itself is separately evidenced · strength of the relation: 1 · note: The paper defines the embedding representation with an i.e. clause. Quotation withheld because this record is not cleared for redistribution. phenomenological models —contrasted-with→ mechanistic models phenomenological models mechanistic models differs in: reproducing empirical data without versus incorporating growth mechanisms · the relation itself is separately evidenced · strength of the relation: 1 · differs in (stated by the tagger, no quotation): reproducing empirical data without versus incorporating growth mechanisms · note: The methods section explicitly distinguishes the two model families. Quotation withheld because this record is not cleared for redistribution. Word embeddings —operationalised-as→ distributed representations of words Word embeddings distributed representations of words the relation itself is separately evidenced · strength of the relation: 1 · note: An i.e. clause defines word embeddings as distributed representations. Quotation withheld because this record is not cleared for redistribution. coefficient of relative amplification —operationalised-as→ E U /E R coefficient of relative amplification E U /E R the relation itself is separately evidenced · strength of the relation: 1 · note: The coefficient is explicitly defined as the unreliable-to-reliable amplification ratio. Quotation withheld because this record is not cleared for redistribution. interaction paradigm imposed by the specific social media —causes→ information spreading interaction paradigm imposed by the specific social media information spreading the relation itself is separately evidenced · strength of the relation: 0.7 · note: The analysis suggests that platform interaction paradigms drive spreading; weight reflects 'suggests'. · the relation stated: information spreading is driven by the interaction paradigm imposed by the specific social media or/and by the specific interaction patterns of groups of users engaged with the topic Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but 'Our analysis suggests' spreading is driven by the interaction paradigm; hedged. Quotation withheld because this record is not cleared for redistribution. the growth of questionable-source content is a re-scaled version of reliable-source content, and the re-scaling factor depends strongly on the platform the growth of questionable-source content is a re-scaled version of reliable-source content, and the re-scaling factor depends strongly on the platform Quotation withheld because this record is not cleared for redistribution. information from reliable and questionable sources does not present different spreading patterns information from reliable and questionable sources does not present different spreading patterns Quotation withheld because this record is not cleared for redistribution. on Gab, questionable posts are about 70% of the reliable volume while their reactions are about three times bigger on Gab, questionable posts are about 70% of the reliable volume while their reactions are about three times bigger Quotation withheld because this record is not cleared for redistribution. reliability is labelled at the level of the news outlet rather than the single article, which the authors name as the limit of the approach reliability is labelled at the level of the news outlet rather than the single article, which the authors name as the limit of the approach Quotation withheld because this record is not cleared for redistribution. the definition of fake news may be inadequate because political debate resorts to labelling opposite news as unreliable the definition of fake news may be inadequate because political debate resorts to labelling opposite news as unreliable Quotation withheld because this record is not cleared for redistribution. information spreading appears driven by the interaction paradigm imposed by the specific platform or by the interaction patterns of the users engaged information spreading appears driven by the interaction paradigm imposed by the specific platform or by the interaction patterns of the users engaged Quotation withheld because this record is not cleared for redistribution. a news outlet systematically showing one or more of: extreme bias, consistent promotion of propaganda or conspiracies, poor or no sourcing, information not supported by evidence or unverifiable, a complete lack of transparency, or fake news questionable sources Quotation withheld because this record is not cleared for redistribution. the average number of reactions to a post - a measure of the extent to which a post is amplified in a social media amplification factor Quotation withheld because this record is not cleared for redistribution. days of data across five platforms, more than 8 million comments and posts 45 Quotation withheld because this record is not cleared for redistribution. R0 R0 Quotation withheld because this record is not cleared for redistribution. platform platform Quotation withheld because this record is not cleared for redistribution. amplification factor amplification factor Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1038/s41598-020-73510-5.