Consolidation Rules

Use these rules when merging lane outputs into final_issues.json.

Merge when

• two findings flag the same underlying defect
• they point to the same quote or the same local text span
• fixing one root cause would resolve both findings

Keep separate when

• the same root cause creates different paper-level consequences
• the findings require different author actions
• one finding is about methodology and the other about claim accuracy
• one uses evidence from a different section or appendix

Priority rules

• keep the better explanation
• keep the higher severity if duplicates disagree
• keep the higher confidence if explanations are otherwise equivalent
• preserve all related sections

False positives

Drop a finding only if it is clearly resolved by context, convention, or parsing noise. Do not drop singleton findings just because they appear only once.