Summary
We've run the auditor against several real published articles but the reference lists and results haven't been committed to the repo. These are needed for ongoing false-positive rate validation.
Known completed audits
The following articles have been audited in previous sessions:
Acceptance Criteria
- Each article's reference list saved as
test-sets/real-articles/{first-author-year}.md
- Table in
test-sets/real-articles/README.md updated with article name, journal, date tested, and result summary
- ≥90% Defensible classification rate for clean articles (anything lower indicates false-positive drift)
Notes
Extract reference lists from the original articles. If the full reference list isn't available from a prior chat session, re-extract from the published PDF.
Summary
We've run the auditor against several real published articles but the reference lists and results haven't been committed to the repo. These are needed for ongoing false-positive rate validation.
Known completed audits
The following articles have been audited in previous sessions:
Acceptance Criteria
test-sets/real-articles/{first-author-year}.mdtest-sets/real-articles/README.mdupdated with article name, journal, date tested, and result summaryNotes
Extract reference lists from the original articles. If the full reference list isn't available from a prior chat session, re-extract from the published PDF.