ICSME Research Reproducibility
Prepares and audits research artifacts, data-availability statements, and experiment provenance for IEEE ICSME and ROSE Festival submissions.
ICSME Research Reproducibility is a specialized skill designed to help software engineering researchers audit paper claims, craft robust data-availability statements, and package verifiable artifacts for the IEEE International Conference on Software Maintenance and Evolution (ICSME) and the ROSE Festival. It provides concrete guidance on provenance pinning across repository mining, LLM prompt caching, and random seeds, ensuring seamless alignment between empirical results in the PDF and underlying execution scripts from double-anonymous review through to camera-ready IEEE badge evaluation.
Key Features
01Maps empirical claims, tables, and metrics directly to verifiable artifact execution scripts
02Pins experiment provenance across Git commit SHAs, cached LLM responses, compute specs, and seeds
03Assesses reproducibility levels across turnkey, scripted, and descriptive classifications
04Audits data-availability statements to eliminate unaccepted 'available on request' promises
051,079 GitHub stars
06Verifies artifact anonymization for double-anonymous review and smooth transition to permanent DOIs
Use Cases
01Documenting prompt provenance, LLM cache records, and nondeterminism bounds in AI-driven SE studies
02Auditing empirical consistency between LaTeX tables and replication packages ahead of ROSE Festival evaluation
03Packaging software mining datasets, analysis scripts, and reproduction guides for ICSME paper submissions