A high similarity score is the start of a conversation, not the end of one. This guide covers how to move from an originality flag to a fair, documented decision — and why keeping a human in the loop protects both the institution and the student.
The module separates four different things on purpose: similarity (text overlaps a source), plagiarism risk (that overlap may be misconduct), AI-assistance signal (later phase), and policy breach (a specific rule was broken). Never treat similarity alone as guilt.
1. Triage from the dashboard
Start on the Integrity Dashboard. The High risk and Review required tiles tell you where to look. Open Originality Checks and filter by risk band to see the queue.
2. Review the originality report
Open a flagged check. Ask:
- Where is the overlap? Matched sources tell you whether it’s another student, a prior submission, or repository material.
- Is it legitimate? Quoted definitions, standard terminology, references, and shared templates naturally produce overlap. A high score on a technical assignment is often not misconduct.
- Is it collusion? Two same-cohort students matching each other is the signal the Similarity Heatmap is built for — load the assessment reference to see the cluster.
3. Apply academic judgement
If the overlap looks legitimate (citations, templates, common phrasing), record your note and clear it. If it looks like misconduct, gather the evidence and move to a case.
4. Open an academic integrity case
skillSYMS records plagiarism and collusion cases in the existing irregularity case system (case_type = plagiarism or cheating). A case captures:
- The student, assessment, and the originality report as evidence.
- Your investigation notes and the student’s explanation.
- Moderator review where your policy threshold requires it.
- The decision (substantiated / unsubstantiated / inconclusive), any sanction, and a full audit trail.
5. Give the student a fair hearing
Before any outcome, the student should be able to see the allegation summary and respond — explain their approach, show drafts, or declare tool use. This is what makes an outcome defensible on appeal.
6. Moderator oversight
For QCTO, SETA, and accredited environments, moderators provide the quality-assurance layer: checking that lecturer decisions are consistent, that high-similarity cases weren’t waved through, and signing off the outcome.
Principles to hold onto
- Similarity is not guilt; investigate before deciding.
- Keep the decision with authorised staff — the system informs, it does not judge.
- Document everything: the report, the notes, the response, the outcome.
- Use the patterns you see to improve assessment design — if a whole cohort matches one source, the question may need a local, individualised prompt.