Skip to main content

Evaluation Workflow

Lumen connects answer-sheet upload, extraction, rubric-based AI suggestions, teacher review, moderation, and final result export.

1. Prepare assessment inputs​

Start with the question paper, model answers or marking scheme, candidate context, and answer sheets. Confirm that the pilot scope is intentionally constrained.

2. Confirm rubrics​

Rubrics define what the AI suggestion looks for. Better rubrics produce more reviewable suggestions.

Rubrics identify:

  • maximum marks
  • required concepts
  • optional concepts
  • common mistakes
  • teacher instructions
  • strictness expectations

3. Extract answer evidence​

Lumen extracts answer evidence from uploaded sheets. Extraction confidence stays visible so reviewers understand where source evidence may be weak.

4. Generate suggested marks​

Lumen generates suggested marks, explanations, missing concepts, and confidence signals against the approved rubric.

5. Review and moderate​

Teachers approve or override suggestions. Moderators handle low-confidence, disputed, or exception cases.

6. Export finalized results​

Only reviewed and approved marks move into exports, reports, or SIS/ERP handoff workflows.