Bind the report to the submission it actually analysed
Check filename, extent, creation date, product version or access state and known settings. Were bibliography, quotations, small matches or certain repositories excluded? Which language and document components were analysed? Mark absent details as unknown instead of supplying a presumed configuration.
Preserve the raw report and annotate a copy. A later run may differ after database or configuration changes. Assign each run a unique ID. Compare outputs only where text state and conditions are traceable. A lower number does not automatically invalidate an earlier report.
Read match types through their attribution function
A properly marked quotation may have extensive identical wording and still comply with the rule. A paraphrase may share few words yet reproduce reasoning without adequate attribution. Bibliographies create characteristic matches, as do statutory titles, standard questionnaires, method names and institutional declarations. Classification therefore comes before judgement.
Tables, figures, code and translations bring additional questions. Is the source named next to the object? Is adaptation declared? Does the licence permit use? Is a translation attributed? A text report cannot detect every visual or structural reuse. Examine the work beyond highlighted prose only where a concrete reason exists.
Several repositories may contain the same passage. A report frequently displays a technically available candidate rather than the source the student used. Trace original publication, edition and page context. An aggregator may itself have copied earlier material. Preserve the source chain instead of making a premature provenance claim.
Do not infer cause or relevance from colour and percentage
A match may be a compliant quotation, an incomplete citation, template wording or a material attribution problem. Highlight colour cannot perform that classification. Consider proximity, extent, location, repetition and importance to the argument or method. A short technical term serves a different function from copied result interpretation.
Intent cannot be read directly from word overlap. The first academic task is often narrower: what attribution is missing, and what assessed contribution becomes unclear? Further conclusions require process records, a response and the competent decision-maker. Use neutral categories while the facts remain under review.
Headline measures are particularly easy to overstate because exclusions, repository coverage and document design affect them. Do not set a universal acceptable percentage. Explain relevance at passage level under the task and rule. Providers themselves distinguish a similarity measure from a plagiarism determination.
Document exclusions and possible problems with equal care
Where a match is a proper quotation or bibliography entry, record the exclusion reason and controlling rule. This shows that the passage was examined rather than ignored. For a possible problem, preserve the submission extract, source, context and open question. Avoid unnecessary reproduction of a third party’s full text.
A student question identifies passage and expectation. For example: “Please explain how paragraph 3.2 developed and its relationship to source X.” It does not assume that deception has been established. Provide access to the material required for an informed response under the institution’s procedure.
The multi-source workflow for educators shows the next review stage. Contrary findings and alternative causes belong in a plagiarism case record too.
Final check before formal referral
- Report and submission are unambiguously connected.
- Known configuration and missing information are separate.
- Every material passage was read in context.
- Source origin was not inferred solely from result order.
- Type, attribution, cause and relevance remain separate fields.
- No headline score operates as an automatic threshold.
- Exclusions and open questions are documented.
- The authorised role receives an unchanged traceable record.
The higher education practice hub connects institutional workflows. The report inventory is explained in requesting a raw similarity report. The one foundation connection is academic writing and evidence.
When several educators review reports, use shared field definitions and explain borderline classifications. A second reader can identify inconsistent treatment without turning agreement into an invented performance statistic. Preserve both observations where disagreement remains.