Recursos · 40

Triagem de sinais de escuta social antes de agir.

Separar volume, origens, republicações e fatos verificados para encaminhar casos úteis.

Atualizado · 6 min

Sala de análise fictícia vazia com painéis de vidro fumado, ligações em latão e caixa bordô Ilustração editorial gerada com IA

O que este guia ajuda a alcançar

  • Descrever a coleta.
  • Desduplicar.
  • Verificar fatos
  • Encaminhar e fechar.

Verificação rápida

  • Quais fontes e idiomas estão faltando?
  • As menções são independentes ou republicações?
  • O que sabemos além da publicação?
  • Que efeito no mundo real justifica a escalada?
  • A resolução foi confirmada?

Método passo a passo

  1. 1

    Defina a pergunta.

    Especifique a decisão, os temas, as plataformas, os idiomas, o período e os limites de coleta. Não trate os participantes online como uma população representativa.

    Entregável: protocolo e cobertura declarada.

  2. 2

    Mantenha a proveniência.

    Registre o link, a data, o contexto visível e o método de descoberta para cada sinal. Retenha apenas os dados pessoais necessários para o tratamento.

    Entregável: registro de sinal datado.

  3. 3

    Elimine repetições duplicadas.

    Agrupe citações, cópias, capturas de tela e publicações subsequentes sobre o mesmo evento. Separe o volume de distribuição de casos independentes.

    Entregável: linha do tempo e clusters de origem.

  4. 4

    Verificar alegações

    Buscar fontes primárias, contexto e evidências contrárias. Marcar o que é confirmado, plausível ou desconhecido, sem inferir intenções a partir de uma única publicação.

    Entregável: status dos fatos e nível de confiança.

  5. 5

    Encaminhar por efeito

    Aplicar critérios de gravidade e atribuir reclamações, riscos de segurança ou rumores à equipe correta. Elaborar respostas que distingam os fatos da investigação em andamento.

    Entregável: responsável, cronograma e decisão.

  6. 6

    Encerrar e aprender

    Confirmar o resultado, corrigir declarações públicas quando necessário e refinar os termos de busca e as regras de triagem. Comparar casos resolvidos, não apenas menções.

    Entregável: encerramento e revisão do método.

Indicadores de gestão

IndicadorO que medePrimeira ação
AbrangênciaFontes e idiomas realmente observadosSinalizar pontos cegos
DesduplicaçãoCasos distintos por cluster de republicaçãoEvitar contagem dupla
QualificaçãoCasos com fatos e confiançaAmpliar a verificação
ResoluçãoCasos escalados efetivamente encerradosRevisar a responsabilidade e o momento

Erros comuns

  • Tratar menções como opinião representativa
  • Contabilizar cada cópia como um novo incidente
  • Publicar conclusões antes da verificação
  • Reter dados pessoais desnecessários

Perguntas frequentes

Um pico de menções comprova uma crise?

Não. Pode vir de uma campanha, de uma coleção alterada ou de repetições de uma mensagem.

Todo sinal deve receber uma resposta?

Isso depende do impacto, da verificação, do canal e do risco de amplificação.

Como a incerteza deve ser demonstrada?

Separar fatos observados, hipóteses e informações ainda em busca de informações.

Cadeias de citações e escuta social

As notas detalhadas desta edição estão disponíveis em inglês.

Useful social listening explains what circulates, which records support a claim and which decision needs preparation. Ten reposts of one document do not become ten confirmations. This dossier adds a way to read documentary dependencies and a briefing structure for OSINT monitoring, reputation and crisis communication.

Official reference: RESIST 3, published 29 October 2025. CIE applications, cases and workshop rules are educational material, not an official RESIST tool or a disinformation diagnosis.

Government Communications · RESIST 3 · RESIST 3 · 2025 · consulted 2026-10-23

Six stages from a signal to a reviewed decision

Six stages from a signal to a reviewed decision
RESIST 3 stageShort referenceOriginal practical application
1. Recognise
Recognise
Separate checkable information, opinion and missing context.CIE application: retain the exact question to be checked before classifying content. Strong criticism does not establish manipulation.
2. Early warning
Early warning
Organise monitoring around defined risks.CIE application: record time window, channels, languages, queries and exclusions. Compare like-for-like periods and document collection outages.
3. Situational insight
Situational insight
Turn observations into decision-relevant information.CIE application: separate message counts, cited documents and observed effects. Identify the question that remains unanswered.
4. Impact analysis
Impact analysis
Prioritise consequences and uncertainties.CIE application: name the exposed service, owner and next review. A rise in reposts alone cannot quantify harm.
5. Strategic communications
Strategic communications
Choose a proportionate response or monitor without responding.CIE application: specify the audience to help, useful fact and suitable channel. Consider whether a public response would increase the content’s visibility.
6. Tracking effectiveness
Tracking effectiveness
Evaluate communication and the response process.CIE application: define an observable outcome before action, then record what actually changes and what remains unknown. Clicks and resolution are different measures.

ABCDE: prepare a briefing without filling unknowns

ABCDE: prepare a briefing without filling unknowns
FieldQuestion to documentPitfall to avoid
ActorWhich accounts or organisations are actually identified?Inferring nationality, intent or coordination from tone.
BehaviourWhich actions were observed, and when?Turning timing similarity into proof of coordination.
ContentWhich exact claim and document version?Treating a headline, cropped image and full text as equivalent records.
DegreeWhat spread was measured, within which scope?Presenting mentions as unique people or representative opinion.
EffectWhat effect was observed, and on whom?Conflating possible effect, observed harm and established causation.

Government Communications · RESIST 3 quick reference · RESIST 3 · 2025 · consulted 2026-10-23

Seguir referências

Choose a small fictional network. An arrow “A → B” means only that A cites B. The workshop follows references reachable from the selected record and shows declared source documents, relays, missing references and cycles. Changing the starting record shows why reading scope matters.

Local calculation on eight closed datasets: no social media collection, transmission, storage or download. “Source” means a declared leaf in the exercise, without external verification. Two distinct identifiers establish neither independence, corroboration nor truth. The result measures neither real audience, a campaign nor an actor’s intent. Displayed loops are those encountered during traversal, not an exhaustive inventory of every possible cycle.

Eight documented readings, available without JavaScript

S01 · Three relays, one document

Documento inicial : D

Three relays, one document
Fictional recordDeclared typeCites records
ADeclared source—
BRelayA
CRelayA
DRelayB, C

D reaches three relays and one leaf A. B and C are not two additional confirmations.

S02 · Two declared documents

Documento inicial : C

Two declared documents
Fictional recordDeclared typeCites records
ADeclared source—
BDeclared source—
CRelayA, B

Two leaves are declared. Their independence and agreement on a particular claim would still require examination of the actual documents.

S03 · Shared origin behind two intermediaries

Documento inicial : D

Shared origin behind two intermediaries
Fictional recordDeclared typeCites records
ADeclared source—
BRelayA
CRelayA
DRelayB, C, A

Three paths reach A. The workshop deduplicates the identifier rather than counting each path as a new source.

S04 · A missing reference

Documento inicial : B

A missing reference
Fictional recordDeclared typeCites records
ADeclared source—
BRelayA, X

A is present, X is absent from the dataset. This does not establish that X is false or nonexistent; it records a reading limitation.

S05 · Two relays cite each other

Documento inicial : A

Two relays cite each other
Fictional recordDeclared typeCites records
ARelayB
BRelayA

The A → B → A cycle provides no declared leaf. Circular repetition does not resolve origin.

S06 · One leaf and a cycle

Documento inicial : B

One leaf and a cycle
Fictional recordDeclared typeCites records
ADeclared source—
BRelayA, C
CRelayB

One leaf A is reachable alongside the B ↔ C cycle. A’s presence does not remove the circular dependency.

S07 · An undocumented origin

Documento inicial : B

An undocumented origin
Fictional recordDeclared typeCites records
AUnknown origin—
BRelayA

A is not declared a source: its origin is unknown. A record without citations must not automatically be promoted to an originating source.

S08 · The starting record changes scope

Documento inicial : B

The starting record changes scope
Fictional recordDeclared typeCites records
ADeclared source—
BRelayA, A
CRelayZ
DUnknown origin—

From B, only B and A are reachable despite the repeated A reference. C and D stay outside the traversal. Selecting C reveals missing reference Z.

Checks before circulating a monitoring brief

Checks before circulating a monitoring brief
CheckExpected record
ScopePeriod, time zone, channels, languages and coverage limits.
Counting unitDistinguish messages, observed accounts, documents, interactions and people.
DependencyTrace citations and keep unknown origins unknown.
VersionCompare publication time, collection time and the version actually read.
Alternative hypothesisCheck editorial reuse, mistakes or a different context before attribution.
DecisionName the action, owner, deadline and new evidence that would change the brief.

For an actionable brief, keep three statements separate: what was observed, what was inferred and what is missing. Link each inference to its records and limitation. Monitoring without a public response can be an explicit decision; give it an owner and a review deadline.

Sources, editions and credits

Original CIE summaries informed by Government Communications, RESIST 3 (2025) and its quick reference. English headings help locate official sections. No institutional logo, diagram or photograph is reproduced. Graph examples are fictional and describe no real person or operation.

Government Communications · RESIST 3 · RESIST 3 · 2025 · consulted 2026-10-23

SHA-256 : 014c25d96b2538b056541179fee24adb086770c8bdf9e382eff055279f01756b

Government Communications · RESIST 3 quick reference · RESIST 3 · 2025 · consulted 2026-10-23

SHA-256 : a27c918181a50443826fb5eb9c140741a0c53fcc4291cc45f4f3d975d82133b1

Referências oficiais

As referências apoiam o método. Adapte as verificações ao seu contexto; elas não são uma certificação. Os títulos das referências originais e os documentos de origem podem estar em outro idioma.

Referências consultadas em .