Turn draft citations into an actionable queue for identity, claim, locator, version, and permission checks. 제공할 내용: Draft claims and citations, Available source details, Publication and verification rules. 예상 결과: A deduplicated source register and prioritized verification checklist with unresolved claim risks.
1
맥락 추가
텍스트는 이 브라우저에 유지됩니다. AILesson Prompts는 이를 모델이나 서버로 보내지 않습니다.
2
프롬프트
채워지지 않은 필드는 플레이스홀더로 표시되므로 프롬프트를 복사하고 편집할 수 있습니다
Turn the supplied citations and sources into a verification checklist. Do not claim to have opened or verified material that is not provided.
Numbered draft claims, citations, quotes, links, and bibliography:
[draft]
Available files, metadata, identifiers, versions, locators, dates, archives, and restrictions:
[sources]
Citation style, evidence standard, deadline, publication status, quote limits, and reviewers:
[rules]
Assign stable Claim and Source IDs. Normalize and deduplicate sources without merging distinct editions, versions, translations, preprints, corrections, datasets, or derivative reports. Map every factual claim and quotation to candidate source, exact locator, version, and support type: direct support, partial support, contextual only, contradiction, or not found. Preserve secondary citation chains and require checking the primary source when the draft implies primary evidence.
Create prioritized checks for source identity/authorship/date, retrievability, version/correction status, claim fidelity, number/unit/denominator, quotation accuracy/context, locator, citation formatting, license/permission, archive/access date, and confidentiality. Each item needs owner or Unassigned, evidence of completion, status, blocker, and deadline. Flag orphan citations, uncited claims, unused bibliography, broken links, circular citations, retractions/corrections needing review, and one source used beyond its scope. End with release gates, a do-not-publish list, and a reconciliation count.
기본적으로 비공개프롬프트 구성은 브라우저에서 로컬로 이루어집니다. 조직에서 허용하지 않는 한 기밀 정보를 AI 서비스에 입력하지 마세요.
입력에서 결과까지
적용 예시
구체적인 맥락이 이 레시피를 바로 사용할 수 있는 결과로 바꾸는 방법을 확인하세요
실제 입력
Draft claims and citations
C1: 'Texts increased reservation pickup by 18%' [Vendor 2025]. C2: A randomized study found 63% versus 58% collection [Chen et al., 2026, Table 2]. C3: Our pilot proves texts work locally [City Library 2026]. Quote Q1: 'reminders are universally effective' (Blog summary). Bibliography also lists Rural Dashboard 2026 but it is not cited.
Available source details
S1 vendor-case.html saved 3 May 2025; page says 'raised pickup by 18%' but no denominator/method, no archive URL. S2 Chen-preprint-v1.pdf DOI pending, Table 2 pages 14-15; repository page says v2 posted 20 Aug 2026 with corrected duplicate records, but v2 file not supplied. S3 city-pilot-final.pdf dated 26 Aug 2026: 61% vs prior-period 54%, explicitly non-randomized, page 6. S4 blog URL accessed 25 Aug cites Chen preprint but provided excerpt does not contain Q1. S5 rural-dashboard.xlsx, Jan-Jun 2026, all ages, no intervention.
Publication and verification rules
Public evidence memo, APA 7, publish 5 Sep 2026. Direct effect claims require primary empirical source. Editor Ana owns copy; analyst Jo checks numbers; legal owner unassigned for quote permission. Maximum short quotation only.
예시 출력
Register: S1 vendor case; S2-v1 Chen preprint; S2-v2 distinct corrected version, metadata only; S3 local final report; S4 secondary blog; S5 rural dashboard. Do not merge S2 versions.
P0 checks before publication: V1/C2—Jo obtains S2-v2, checks Table 2 values, denominators, duplicate correction, pages, DOI/version, and records screenshot or note; status Blocked because v2 absent. Current v1 cannot support final numbers after known correction. V2/C3—Ana changes “proves” to a descriptive statement: local pilot was 61% versus 54% in a non-randomized prior-period comparison (S3 p.6); direct support after wording change. V3/Q1—remove from publishable draft unless exact words, primary locator, context, and permission are verified. S4 excerpt does not contain it, and “universal” exceeds supplied evidence.
P1: C1 is directly quoted from S1 but the source provides no method/denominator, so it supports only “vendor reports 18%,” not an effect claim. Either label and contextualize or omit; archive URL and relative/point meaning remain Unassigned. C2 secondary chain via S4 should cite verified S2-v2 primary, not the blog. S5 is unused bibliography and non-comparable; remove unless used as clearly labeled context.
Release gates: no unsupported Q1; no causal C3; no C2 number until corrected primary version checked; all public URLs archived/access-dated; APA metadata complete; confidentiality clear. Do-not-publish now: Q1, original C3, definitive C1, current C2 number. Reconciliation: 4 draft claims/quotes mapped, 5 logical source records plus one version record; 1 unused source; 0 verified-complete items from supplied files because v2 and permissions remain open.
효과가 있는 이유
1
Claim-source mapping tests whether a citation supports the nearby statement rather than merely discussing the topic
2
Version and primary-source checks prevent a polished bibliography from hiding obsolete or derivative evidence
결과 확인
Is every factual claim and quotation mapped to a source, locator, version, and support status
Are distinct editions, corrections, datasets, and secondary citation chains preserved
Do release gates prevent publication while material verification or permission remains open
안심하고 사용하세요
자주 묻는 질문
이 레시피를 언제 사용해야 하는지, 무엇을 제공해야 하는지, 그리고 어떤 부분에서 사람의 검토가 여전히 중요한지에 대한 실용적인 답변
What should I prepare before using “Organize citations and sources into a verification checklist”?
For “Organize citations and sources into a verification checklist,” prepare Draft claims and citations, Available source details, and Publication and verification rules. Replace placeholders only with information you can verify. If a detail is unknown, preserve that uncertainty explicitly instead of asking the model to infer it.
When is the “Organize citations and sources into a verification checklist” result not ready to use?
The result is not ready if it does not yet deliver the stated outcome—A deduplicated source register and prioritized verification checklist with unresolved claim risks—from the supplied evidence, or if it relies on unresolved assumptions, missing approvals, or invented details. Use the checks as release gates: revise the source inputs or assign a named, authorized reviewer instead of polishing an unsupported output.
Which AI tools have recorded tests for “Organize citations and sources into a verification checklist”?
The published test record for “Organize citations and sources into a verification checklist” lists ChatGPT as of 2026-08-28. This confirms recorded runs, not guaranteed compatibility or identical results in later product versions. For another tool or version, keep every constraint visible and repeat the result checks before use.
ChatGPT
