paper-claim-audit

To verify quantitative claims against raw result files, you need to audit paper .tex source files and raw JSON/CSV/YAML data to eliminate confirmation bias. This pre-submission check catches number inflation, cherry-picking, and aggregation errors by using a fresh zero-context cross-model reviewer to produce a structured per-claim verification report.

A zero-context fresh reviewer audit uses a separate, context-free cross-model reviewer to compare paper claims against raw experimental data. This mechanism eliminates executor confirmation bias by independently verifying all numbers, percentages, and scope statements without prior exposure to the research context.

Checking for best-seed cherry-picking and number inflation requires comparing all reported metrics in your .tex source files against raw experimental result files. A comprehensive pre-submission audit verifies every number, percentage, and comparison to ensure exactly matched data and flags configuration mismatches.

Yes, you need paper .tex source files and raw result JSON/CSV/YAML files to perform a paper-to-data mismatch audit. Access to these specific formats allows the reviewer to parse quantitative claims and exactly match them against raw experimental outputs to eliminate data misreporting.

The best way to audit scope overclaims and configuration mismatches for MICCAI submissions is running a comprehensive pre-submission check that verifies all scope statements, figure captions, and table data against raw result files. This process outputs a structured audit report with per-claim verification status and an overall pass, warn, or fail verdict.

A pre-submission paper audit catches data misreporting including number inflation, best-seed cherry-picking, configuration mismatches, aggregation errors, delta errors, and scope overclaims. It verifies all numbers, percentages, comparisons, and figure captions against raw experimental result files to prevent submission rejections.