참고 자료 · 40

조치 전에 소셜 리스닝 신호를 분류합니다.

유용한 사례를 분류하기 위해 볼륨, 출처, 재게시 및 검증된 사실을 분리합니다.

업데이트됨 · 12 min

스모크 유리 패널, 황동 연결부와 버건디 문서함이 있는 가상의 무인 분석실 AI로 생성한 편집용 일러스트

이 가이드가 달성하는 목표

  • 수집 설명
  • 중복 제거
  • 사실 확인
  • 라우팅 및 종료

빠른 확인

  • 누락된 소스와 언어는 무엇입니까?
  • 언급은 독립적인 게시물인가요, 아니면 재게시물인가요?
  • 게시물 외에 추가로 알 수 있는 정보는 무엇인가요?
  • 어떤 실제적인 영향이 보고를 필요로 하나요?
  • 해결이 확인되었나요?

단계별 방법

  1. 1

    질문을 정의하세요.

    결정, 주제, 플랫폼, 언어, 기간 및 수집 제한 사항을 명시하세요. 온라인 참여자를 대표 집단으로 간주하지 마세요.

    결과물: 프로토콜 및 명시된 적용 범위.

  2. 2

    출처를 유지하세요.

    각 신호에 대한 링크, 날짜, 보이는 맥락 및 발견 방법을 기록하세요. 처리에 필요한 개인 정보만 보관하세요.

    결과물: 날짜가 포함된 신호 기록.

  3. 3

    중복 제거.

    동일한 이벤트에 대한 인용문, 사본, 스크린샷 및 후속 게시물을 그룹화하세요. 배포량과 독립적인 사례를 구분하세요.

    결과물: 타임라인 및 출처 클러스터.

  4. 4

    주장 검증

    1차 자료, 맥락 및 반대 증거를 찾습니다. 하나의 게시물에서 의도를 추론하지 않고 확인된 내용, 그럴듯한 내용 또는 알려지지 않은 내용을 표시합니다.

    결과물: 사실 상태 및 신뢰도.

  5. 5

    영향에 따른 분류

    심각도 기준을 적용하여 불만 사항, 보안 위험 또는 소문을 적절한 팀에 배정합니다. 사실과 진행 중인 조사를 구분하는 답변을 작성합니다.

    결과물: 담당자, 시기 및 결정.

  6. 6

    종결 및 학습

    결과를 확인하고 필요한 경우 공개 성명을 수정하고 검색어 및 분류 규칙을 개선합니다. 언급된 사례뿐 아니라 해결된 사례도 비교합니다.

    결과물: 종결 및 방법 검토.

관리 지표

지표측정 대상첫 번째 조치
범위실제로 관찰된 출처 및 언어사각지대 표시
중복 제거재게시 클러스터별 고유 사례중복 계산 방지
자격 검증사실과 확신이 있는 사례검증 확장
해결에스컬레이션된 사례의 실제 종결소유권 및 시점 검토

일반적인 실수

  • 언급을 대표 의견으로 간주
  • 각 사본을 새로운 사건으로 계산
  • 확인 전 결론 게시
  • 불필요한 개인 데이터 보존

자주 묻는 질문

언급 급증이 위기를 의미하는가?

아니요. 캠페인, 변경된 데이터 수집 또는 동일한 메시지의 반복 게시로 인해 발생할 수 있습니다.

모든 신호에 응답해야 하는가?

영향, 검증, 채널 및 증폭 위험에 따라 다릅니다.

불확실성은 어떻게 표시해야 합니까?

관찰된 사실, 가설 및 아직 수집 중인 정보를 구분합니다.

인용 사슬과 소셜 모니터링

이 언어판의 상세 참고 설명은 영어로 제공됩니다.

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

참조 추적

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

시작 문서 : 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

시작 문서 : 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

시작 문서 : 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

시작 문서 : 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

시작 문서 : 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

시작 문서 : 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

시작 문서 : 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

시작 문서 : 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

공식 참조

참조는 방법을 뒷받침합니다. 상황에 맞게 검사를 조정하십시오. 이는 인증이 아닙니다. 원본 참조 제목 및 소스 문서는 다른 언어로 작성되었을 수 있습니다.

참고 자료 확인일 .