What the report is flagging
Common hits: stacked related work, methods/experimental boilerplate, definitions, your own preprints or papers, conference–journal overlap, web pages and tech reports.
Databases differ—percentages are not comparable across tools. Editors and graduate schools care about uncited overlap and duplicate publication, not winning a score contest.
A sensible order of fixes
Cite and paraphrase properly first; then rewrite needless near-copies. For your own prior text, disclose and use allowed exclusions—do not only synonym-swap to fool the checker.
For extensions and theses, name prior versions in the letter and text. A “clean” score with undeclared reuse is riskier than an honest overlap note.
- Triage each highlight: missing cite vs rewrite vs legitimate reuse
- Compress methods by citing standard procedures instead of pasting blocks
- Keep necessary terms and equations—do not scramble notation to evade detection
Talking to editors or the graduate school
If the tool hits reference lists, funding boilerplate, or standard ethics text, attach a short exclusion note. For real overlap, list prior publications and the delta.
After a return, fix the substance—do not only rerun another checker on the same PDF.
FAQ
- What percentage is “safe”?
- No global cutoff—schools and journals set their own. Below-threshold uncited copying can still be misconduct.
- Will reusing my conference paper count as plagiarism?
- Disclosure and duplicate-publication policy matter. Declared extensions are discussable; hidden reuse is worse.