AAAI Artifact Evaluation
Use this to prepare artifacts that reviewers can use to assess reproducibility. AAAI supplementary
material is part of the submission record; after review starts, do not assume it can be updated.
Artifact package
- Provide a technical appendix for proofs, algorithms, assumptions, hyperparameters, and extended
experiments.
- Provide code/data ZIPs that reproduce main tables or figures, with a short README, environment,
commands, seeds, expected outputs, and runtime.
- Provide multimedia appendices only when they support the technical claim.
- Remove author names, usernames, paths, repository history, cloud buckets, API keys, and metadata.
- Avoid web pointers in the reviewed submission unless current rules explicitly allow them.
- Include licensing and access notes for datasets, models, and third-party code.
AAAI-specific discipline
- Treat the supplementary deadline as final.
- Verify ZIP integrity before submission; missing or corrupted files may not be fixable during
rebuttal.
- Make the reproducibility checklist consistent with the artifact package.
- Prepare a post-acceptance public release path but keep review artifacts anonymous.
What an AAAI reviewer actually opens
AAAI does not run a separate badged artifact-evaluation committee the way some systems venues do;
the same broad-AI reviewer who scores the paper also inspects whatever supplement you attach. That
reviewer may be a planning, knowledge-representation, or constraint-satisfaction specialist rather
than a deep-learning engineer, so the artifact has to be legible without insider tooling. Optimize
for a reviewer who skims, not one who will spend an afternoon configuring a cluster.
| Reviewer action |
Passes |
Fails |
| Opens the ZIP |
sane tree, top README |
nested archives, 0-byte files |
| Reads appendix |
maps to numbered claims |
contradicts the paper |
| Tries one command |
reproduces one headline number |
needs private data or credentials |
| Scans for identity |
nothing reveals authors |
Git logs or home paths leak |
Phase-1 artifact red flags
Because clearly-below-bar papers can be cut before author feedback, a supplement that looks thin or
unrunnable is a cheap reason to summary-reject. Avoid these:
- Checklist promises released code, but the ZIP only holds figures and no scripts.
- A "see our repository" pointer to a mutable, deanonymizing URL.
- Multimedia attached for spectacle that carries no technical claim, inflating size with no rigor.
- Datasets shipped with no license note, leaving reuse legality unverifiable.
Worked vignette
A constraint-solving paper claims a 30% node-expansion reduction. The team ships a large ZIP of raw
solver logs but no driver script. The reproduction path is empty, so artifact status is "risky"; the
fix is a small run_main.py that regenerates Table 2 from seeds, a trimmed log sample, and a license
for the benchmark instances. The raw dump moves to the post-acceptance release.
Output format
[Artifact status] complete / partial / risky / unavailable
[Submitted files] technical appendix / multimedia appendix / code-data ZIP
[Reviewer reproduction path] <commands and expected output>
[Anonymity risks] <metadata, links, paths, logs>
[Missing items] <data, code, seeds, licenses, hardware>
1---2name: aaai-artifact-evaluation3description: Use when packaging AAAI code, data, multimedia appendices, technical appendices, reproducibility evidence, and post-acceptance artifact releases without violating double-blind or immutable-supplement rules.4---56# AAAI Artifact Evaluation78Use this to prepare artifacts that reviewers can use to assess reproducibility. AAAI supplementary9material is part of the submission record; after review starts, do not assume it can be updated.1011## Artifact package1213- Provide a technical appendix for proofs, algorithms, assumptions, hyperparameters, and extended14 experiments.15- Provide code/data ZIPs that reproduce main tables or figures, with a short README, environment,16 commands, seeds, expected outputs, and runtime.17- Provide multimedia appendices only when they support the technical claim.18- Remove author names, usernames, paths, repository history, cloud buckets, API keys, and metadata.19- Avoid web pointers in the reviewed submission unless current rules explicitly allow them.20- Include licensing and access notes for datasets, models, and third-party code.2122## AAAI-specific discipline2324- Treat the supplementary deadline as final.25- Verify ZIP integrity before submission; missing or corrupted files may not be fixable during26 rebuttal.27- Make the reproducibility checklist consistent with the artifact package.28- Prepare a post-acceptance public release path but keep review artifacts anonymous.2930## What an AAAI reviewer actually opens3132AAAI does not run a separate badged artifact-evaluation committee the way some systems venues do;33the same broad-AI reviewer who scores the paper also inspects whatever supplement you attach. That34reviewer may be a planning, knowledge-representation, or constraint-satisfaction specialist rather35than a deep-learning engineer, so the artifact has to be legible without insider tooling. Optimize36for a reviewer who skims, not one who will spend an afternoon configuring a cluster.3738| Reviewer action | Passes | Fails |39| --- | --- | --- |40| Opens the ZIP | sane tree, top README | nested archives, 0-byte files |41| Reads appendix | maps to numbered claims | contradicts the paper |42| Tries one command | reproduces one headline number | needs private data or credentials |43| Scans for identity | nothing reveals authors | Git logs or home paths leak |4445## Phase-1 artifact red flags4647Because clearly-below-bar papers can be cut before author feedback, a supplement that looks thin or48unrunnable is a cheap reason to summary-reject. Avoid these:4950- Checklist promises released code, but the ZIP only holds figures and no scripts.51- A "see our repository" pointer to a mutable, deanonymizing URL.52- Multimedia attached for spectacle that carries no technical claim, inflating size with no rigor.53- Datasets shipped with no license note, leaving reuse legality unverifiable.5455## Worked vignette5657A constraint-solving paper claims a 30% node-expansion reduction. The team ships a large ZIP of raw58solver logs but no driver script. The reproduction path is empty, so artifact status is "risky"; the59fix is a small `run_main.py` that regenerates Table 2 from seeds, a trimmed log sample, and a license60for the benchmark instances. The raw dump moves to the post-acceptance release.6162## Output format6364```text65[Artifact status] complete / partial / risky / unavailable66[Submitted files] technical appendix / multimedia appendix / code-data ZIP67[Reviewer reproduction path] <commands and expected output>68[Anonymity risks] <metadata, links, paths, logs>69[Missing items] <data, code, seeds, licenses, hardware>70```71
brycewang-stanford