recsys-workflow

RecSys Workflow

Use this as the project-management skill for a RecSys submission. Replace all dates with the

current official timetable and work backwards from the submission-system cutoffs.

RecSys is a conference, not a journal: it has no standing editor-in-chief and no article-processing

charge. The rotating leadership is the per-edition General Chairs and Program Chairs (2026 General

Chairs: Joseph A. Konstan, George Karypis, Gediminas Adomavicius; Program Chairs: Minmin Chen, Bart

Goethals, Martijn Willemsen — verified 2026-07-09), and the cost model is registration fees, with

proceedings in the ACM Digital Library. Chairs rotate yearly, so re-check the current committees

page rather than carrying a name forward.

Per-track dates are not the same

The 2026 cycle ran several tracks on different timetables — do not assume the Main Track dates

apply to your track.

| Track (2026, AoE) | Abstract | Paper | Notification | Camera-ready |

|---|---|---|---|---|

| Main (long/short/PPF) | Apr 14 | Apr 21 | Jul 9 | 待核实 |

| Reproducibility | Apr 28 | May 5 | Jul 9 | Jul 27 |

| Industry | (site Mar 2) | May 21 | Jul 9 | Jul 27 |

| Resource / Dataset | Apr 28 | May 5 | Jul 9 | 待核实 |

| R&P Notes / Demos | — | Jul 15 | — | — |

The Main-Track rebuttal window was June 4-9, 2026. Anchor every one of these to the current CFP;

they are historical anchors, not permanent rules.

Milestones

• Venue fit: confirm the contribution is genuinely a recommendation result (see

recsys-topic-selection).

• Evidence lock: freeze datasets, the split protocol, tuned baselines, seeds, ablations, and the

anonymous repository contents.

• Abstract deadline: submit real title, abstract, authors, and topic areas.
• Paper deadline: upload the anonymous PDF, the in-paper anonymous repository link, and required

fields.

• Rebuttal: draft a concise anonymous narrative on the central evaluation objection.
• Decision: archive reviews, rebuttal, and the submitted version.
• Acceptance: prepare the ACM camera-ready, rights form, registration, presentation, and the

public artifact release.

Backward plan from the paper deadline

| Weeks out (heuristic) | Offline-evaluation recommender milestone |

|---|---|

| 8+ | Datasets, versions, and split protocol frozen |

| 6 | Baselines tuned under an equal budget; grids logged |

| 4 | Ablations and the off-policy/simulator bridge complete, seeds logged |

| 3 | Full draft in the ACM two-column template |

| 2 | Internal mock review by a reproducibility-minded reader |

| 1 | Anonymity sweep; anonymous repository builds on a clean machine |

| 0 | Abstract then full paper on the submission system |

These offsets are planning heuristics only — anchor them to the current official timetable.

Failure modes by stage

• Baselines still on default configs at week 3 forces an unfair-comparison reject in the making.
• Leaving template conversion to the final week surfaces two-column overflow late.
• Skipping the mock review forfeits the only chance to hear "this split leaks the future" from a

friend instead of a reviewer.

• Building the anonymous repository under deadline pressure is how commit-author leaks ship.

Coordination notes

• Assign one owner for the tuning-and-seeds protocol and another for the anonymity sweep.
• Archive the exact submitted PDF and repository state, since the rebuttal must quote them.

Output format

[Current stage] idea / experiments / writing / abstract / submission / rebuttal / accepted
[Track + next deadline] <track> - <date and source, or unknown>
[Critical path] <three tasks that determine readiness>
[Risk register] <baseline tuning / split / anonymity / repository / presentation>
[Owner map] <task -> person or role>