Ijcai supplementary

IJCAI Supplementary

Use this when assembling or auditing IJCAI supplementary material. Reopen the current CFP and FAQ because deadlines, file types, and size limits are cycle-specific.

Supplement plan

• Separate the Technical Appendix from the Resubmission File. Do not mix prior-review material into the technical supplement.
• Keep the main paper self-contained; reviewers may choose not to read the appendix.
• Submit supplement and resubmission files by the full-paper deadline. IJCAI-ECAI 2026 did not use separate later deadlines for these files.
• Respect file type and size rules. IJCAI-ECAI 2026 allowed up to 50MB of supplementary material in PDF or ZIP format.
• Anonymize every file, archive name, directory name, path, metadata field, license header, screenshot, notebook output, and URL.
• Verify that ZIPs open, contain deterministic paths, and do not include cache files, hidden OS files, credentials, API keys, or large irrelevant artifacts.

Placement decision: body, appendix, or ZIP

IJCAI reviewers may never open the supplement, so placement is a scoring decision. Sort each artifact before assembling.

Artifact	Where it belongs	Why
Theorem statement, main result table	7-page body	Load-bearing; reviewers must see it without the supplement
Full proof, extra derivations	Technical Appendix	Supports credibility but not required reading
Hyperparameter grids, full per-domain tables	Technical Appendix	Detail that would crowd the body
Runnable code, simulator, instance generator	Code/data ZIP	Reproducibility evidence reviewers may sample
Prior reviews and rejected version	Resubmission File only	Must stay out of the technical supplement

Worked vignette: a planning paper's split

A planning paper has a soundness theorem, a 40-domain coverage table, and a planner implementation. Decision: the theorem statement and a condensed coverage summary stay in the body; the proof and the full 40-domain table go to the Technical Appendix; the planner plus a deterministic instance generator and seeds go in the ZIP; nothing from a prior IJCAI rejection goes anywhere except the Resubmission File. This keeps the paper self-contained while giving a skeptical search/planning reviewer a re-runnable artifact.

Reviewer pushback and the venue-specific fix

• "Had to open the ZIP to understand the method." Move the missing protocol into the body; the appendix should deepen, not define, the contribution.
• "Appendix mixes in old reviews." Separate them into the Resubmission File; mixing risks an anonymity or policy flag.
• "Archive would not open or had absolute paths." Re-zip with relative deterministic paths and strip OS and cache files before the full-paper deadline, since there is no later upload.

Output format

[Supplement status] Ready / Needs fixes / Not ready
[Files] <technical appendix / code-data zip / resubmission file>
[Deadline coupling] <same as full paper? source needed>
[Anonymity checks] <passed/issues>
[Reviewer dependency] <what breaks if supplement is ignored>