Ask for the actual record, not a more favourable rerun
Frame the request neutrally: you need to understand which file was checked, by which system and under which conditions, and which passages support the stated concern. Do not begin by demanding a new scan with altered exclusions because it may yield a lower percentage. First obtain the report that was genuinely used in the procedure.
Identify the case or work, submission date and filename. Ask for an accessible format through the approved route while protecting the response deadline. Entitlement to particular records depends on the specific procedure. Where a report contains another person’s data or protected source material, supervised inspection and provision of a copy may be treated differently.
Do not confuse document score, source score and passage match
A document-wide score aggregates similarities according to the product’s rules. It is not an automatic percentage of plagiarism. A source score might arise from many small legitimate references or one substantial unattributed passage. Bibliographies, quotations, assignment wording and standard methods may be included or excluded depending on configuration.
Record the provider’s label and its documented meaning. Do not import a definition from another tool. Check whether percentages overlap, because one passage can resemble several sources. Never add source figures yourself. A rerun at a later date may differ after database or product changes.
The report is a search and review aid. A high number cannot replace assessment of the relevant passages; a low number cannot rule out a particular concern. Attribution, wording, extent, context and the applicable academic rule require separate attention.
Open each comparison source in its real context
Verify title, author, edition, publication date and exact location. For a web source, an archived state may matter when content changed. An aggregator may reproduce material from an older original, so the first displayed candidate is not necessarily the source actually used. Trace the chain without prematurely declaring the report wrong.
Place the submitted passage beside enough surrounding source text to understand both. Record quotation marks, footnotes and bibliography entries that are present. For paraphrase, examine attribution and intellectual sequence as well as shared wording. Technical terms, legislation, formulae and standard instruments need discipline-specific treatment.
If the comparison cannot be accessed, record that limitation and request an inspectable reference or controlled access. Do not reconstruct a whole source from a short highlighted extract. The guide to checking method errors in a similarity report covers exclusions and source misattribution.
Separate machine mark, human review and formal concern
Create a passage register with IDs. The first column reproduces the system mark without alteration. The second records how a human reviewer classified it, where documented. The third gives your observation and evidence. This reveals whether a match was excluded, left unresolved or actually relied on in the allegation.
Keep the raw report unchanged and annotate a copy. If access is portal-only, note the viewing date and permitted export route. Do not publish the report or source files: they can contain copyrighted work, confidential submissions and personal data.
The examination procedures hub connects the report to later steps. For file preparation, see reviewing a plagiarism case file. The one foundation link is academic writing and evidence.
Seven checks before drafting a response
- The report is unambiguously connected to the submitted file.
- Date, product and known settings are recorded.
- Every measure uses the relevant provider definition.
- Matches can be located with surrounding context.
- Comparison sources are identified in the version used.
- Automated marking and human assessment remain distinct.
- Missing components are described as missing.
Preserve your request, the supplied report and the date of access. If another version appears later, assign a new identifier and compare it with the first rather than silently replacing the procedural record.