Skills Data Science Achieving Scientific Reproducibility and Provenance

Achieving Scientific Reproducibility and Provenance

v20260724
icsme-reproducibility
A comprehensive guide for researchers submitting academic papers to ensure high standards of reproducibility and verifiable evidence. It covers critical areas such as data-availability statements, artifact management, provenance pinning (for LLMs and mining), and mapping claims directly to evidence, adhering to rigorous academic standards.
Get Skill
376 downloads
Overview

ICSME Reproducibility

Use this before submission and again before camera-ready. ICSME's ROSE-Festival culture makes reproducibility a valued, sometimes scored dimension rather than a courtesy: double-anonymous review already expects an inspectable artifact, and the Joint Artifact Evaluation Track and ROSE Festival rewards a permanent, open one with IEEE badges. The goal is that a competent maintainer could rebuild your evidence — re-mine the corpus, re-run the technique — and reach your conclusions.

Evidence map

  • Map each research-question answer, technique claim, and reported number to a verifiable location — a paper section, a table generated from logged data, or a script in the artifact.
  • For techniques, give enough of the algorithm, parameters, and environment that a reader could re-implement or re-run it on their own systems.
  • For mining/evolution studies, report subject systems and their selection, the extraction pipeline, preprocessing, metrics, statistics, and the analysis scripts.
  • Keep the data-availability statement truthful and specific: what is shared (corpus, scripts, subject list), where it will live after acceptance, and — if something cannot be shared (proprietary legacy code, private history) — exactly why.
  • Keep the paper and the artifact consistent: a number in the PDF that no script in the artifact produces is the contradiction reviewers read as carelessness.

Data-availability statement audit

Claim in the paper Weak availability answer ICSME-ready answer
"We mine N evolving projects" "Dataset available on request" Anonymized archive of the exact project list, SHAs, and extraction scripts
"Our tool detects debt hotspots" "Code will be released" Anonymized, runnable tool with a README and a small demo on a bundled repo
"We surveyed/interviewed P maintainers" Nothing (privacy cited vaguely) Anonymized codebook, protocol, and aggregate data; stated ethics limits
"The model summarized the change" Live API described Cached prompts and raw responses, model IDs and access dates

"Available on request" is treated as not available under ROSE norms; convert every such line into a concrete, anonymized artifact or an explicit, justified exception.

Provenance pinning

[Mining]   pin repository SHAs; record extraction date and the studied time window; archive the
           extracted corpus, not just the query; document fork/duplicate/bot handling
[LLM]      record exact model identifiers + access dates; cache raw inputs and outputs; report
           sampling settings; prefer post-cutoff subjects to bound contamination
[Compute]  state hardware, mining/runtime cost, and number of runs so a reader can size a reproduction
[Randomness] log seeds for any stochastic step; say what is and is not deterministic

Degrees of reproducibility (state the one you achieved)

  • Turnkey: one documented command regenerates each table/figure from logged data.
  • Scripted: scripts exist but require documented manual steps or access to a large mined corpus.
  • Descriptive: prose detailed enough that a competent reader could rebuild the pipeline.

For ICSME, aim turnkey for anything a reviewer might rerun quickly (a detection script on a sample repo, a plot from logged results); a large mined corpus or proprietary history may stay scripted with access clearly documented. Stating the achieved level honestly beats promising turnkey behaviour that fails on a clean machine — especially with no revision round to fix it in.

Vignette: a mixed-methods evolution study

Consider a study combining mined release history with a maintainer survey. Its reproducibility spine: the mining scripts with pinned SHAs and extraction dates; the anonymized extracted dataset with the exact subject-system list; the survey instrument and anonymized responses; the qualitative codebook with inter-rater agreement; and the analysis notebooks that turn all of it into the paper's tables — plus one honest sentence about the parts (identities, private forks) that cannot be shared and why.

Consistency and camera-ready / ROSE pass

  • Before submission: every scored number traces to the artifact; the data-availability statement matches reality; the artifact is anonymized (no owner strings, cluster paths, lab names, or your own org's repos).
  • Before camera-ready: swap anonymized links for a permanent, DOI-issuing archive, and align the statement with the IEEE badges you pursue in the ROSE Festival (icsme-artifact-evaluation).

Output format

[Claim inventory] <claim -> evidence location>
[Data availability] concrete / vague / missing
[Provenance gaps] <mining SHAs / LLM caching / seeds / compute>
[Reproducibility level] turnkey / scripted / descriptive, stated honestly
[Paper fixes] <must appear in the PDF, since there is no revision round>
[Artifact fixes] <additions before upload>
Info
Category Data Science
Name icsme-reproducibility
Version v20260724
Size 5.26KB
Updated At 2026-07-28
Language