ijcai-review-process

IJCAI Review Process

Use this to reason about review-stage strategy. Reopen the current CFP, FAQ, peer-review principles, and conflict policy before making process claims.

Process model

• Phase 1 can be decisive. IJCAI-ECAI 2026 gave every paper two PC reviews first; papers with two good-quality but insufficiently positive reviews could receive summary rejection without author response.
• Papers advancing to full review receive additional review and discussion by PC/SPC members, with AC coordination and SAC cross-checking.
• The response is concise and visible to the decision group specified by current rules; it is not a dialogue with reviewers.
• Ethics concerns can be flagged by reviewers and escalated to the Ethics Chair.
• Conflicts include recent coauthors, advisor/student relationships, same institution, current grants/projects, sponsorship, personal relationships, or animosity under the IJCAI COI policy.
• Authors must not try to identify or contact reviewers, and reviewers must preserve submission confidentiality.

Decision actors and where leverage lives

Stage	Who decides	Author leverage	Forbidden
Phase 1 summary reject	Two PC reviewers, AC	None: there is usually no response here	Treating phase 1 as if a rebuttal will save it
Full review	PC, SPC, AC discussion	Concise response on pressing questions and factual errors	New experiments, code links, or scope changes
Decision	AC recommends, SAC cross-checks, chairs finalize	Clarity already in the paper, ethics flags resolved	Contacting or guessing reviewer identity
Ethics escalation	Ethics Chair	Confidential channel for unethical-review concerns	Using ethics box to argue scores

Because IJCAI spans many AI subcommunities, a paper can draw one in-area and one out-of-area reviewer; the out-of-area reviewer is often where summary-reject risk concentrates, so the phase-1 readability of the contribution matters more than at single-topic venues.

Worked vignette: surviving phase 1

A constraint-reasoning paper gets two competent reviews that are lukewarm because the second reviewer, from a learning background, could not locate the contribution by page two. Under IJCAI rules this profile is a summary-reject candidate before any response. The leverage is not in the rebuttal: it is in writing so a non-specialist PC member sees the novel claim on page one. Diagnose this risk pre-submission, since phase 1 may give no chance to reply.

Reviewer pushback and the venue-specific fix

• "Reviewer misread the method." If it is decision-critical, correct it precisely in the response with a pointer to the submitted section; do not relitigate the score.
• "Two okay-but-tepid reviews." Recognize this as the classic IJCAI summary-reject pattern and fix readability upstream; you cannot count on a response stage.
• "Suspected reviewer is a competitor." Do not investigate; raise a conflict or ethics concern through the official channel only.

Output format

[Current stage] submitted / phase 1 / full review / response / decision
[Decision actors] <PC/SPC/AC/SAC/chairs>
[Likely leverage] <paper clarity / factual correction / ethics / reproducibility>
[Forbidden moves] <contact reviewers / new results / identity leakage>