brycewang-stanford EDBT Review Process
Model the pipeline before interpreting any single review. EDBT's process is a multiple-cycle rolling model with a real in-cycle revise-and-resubmit: within one cycle, a paper is reviewed, gets an author-feedback exchange, receives accept / revise / reject, and — if revise — is revised and re-read by the same reviewers before a final accept/reject. The most consequential mental shift for authors arriving from a one-shot conference is that a revise is a genuine second chance inside the same cycle, not a soft rejection.
Process model
- Reviewing runs in Microsoft CMT (Conference Management Toolkit). Confirm the blind policy for the cycle — single- vs double-blind is decided per edition (待核实; EDBT has historically used author-identified single-blind reviewing).
- Each paper is read by multiple program-committee members who weigh the significance of the data-management contribution, technical soundness, the fairness and honesty of the evaluation (real workloads, tuned baselines, realistic scale), reproducibility, and clarity.
- The cycle includes an author-feedback phase on the initial reviews, then a first notification of accept / revise / reject.
- A revise becomes a revised submission plus a change letter, re-read — generally by the original reviewers — for a second notification of accept / reject within the cycle.
- Accepted papers publish open access on OpenProceedings.org (CC-BY-NC-ND). Earlier-cycle acceptances present at that year's conference; last-cycle acceptances roll to the next edition.
Reading a decision against the categories
| Decision | What it means | Author move |
|---|---|---|
| Accept | Contribution and evaluation hold; minor polish only | Camera-ready + artifact; do not reopen scope |
| Revise | Repairable gaps: a missing baseline, an unconvincing scale, an unclear mechanism | Treat as an in-cycle R&R: make or explicitly decline every request, evidenced |
| Reject | Structural: wrong problem framing, unfair or missing evaluation, thin contribution | Reframe or reroute (SIGMOD/VLDB/ICDE/ICDT or a journal); remember the 12-month EDBT ban |
The strategic reading: write the initial submission so that whatever is weakest is fixable inside the in-cycle revision window (a baseline you can add, a scale you can extend, a mechanism you can clarify) rather than structural (an evaluation you cannot redo in the cycle). The process rewards repairable papers.
How EDBT differs from its siblings
- vs. SIGMOD / ICDE: those publish in the paywalled ACM DL / IEEE Xplore and run their own cycle structures; EDBT's identity is the OpenProceedings open-access record and the three-cycle rolling cadence. Never assume a shared calendar or template.
- vs. VLDB / PVLDB: PVLDB is a journal-style monthly pipeline with its own revision mechanics; EDBT's revise-and-resubmit is bounded inside a discrete cycle, not a rolling monthly journal.
- vs. ICDT (the co-located sibling): ICDT reviews theory on its own call and PC; sharing a conference week and the OpenProceedings platform does not make them one process. An EDBT paper is judged by DB-systems reviewers, not theory reviewers.
Who reads you
Expect several database-systems reviewers. They look for a real data-management problem, a mechanism described precisely enough to reimplement, an evaluation on real workloads at honest scale against tuned baselines, and a reproducible artifact. For an Experiments & Analysis paper the bar is higher on the measurement itself: methodology, coverage, and repeatability are the contribution.
Where author leverage actually exists
[Before submission] subject-area tags + cycle choice -> reviewer pool + timing (largest lever)
[Author feedback] factual corrections, targeted clarifications, misread fixes
[Revise] the strongest lever: a revised paper + point-by-point change letter,
re-read by the same reviewers inside the cycle
[After reject] no appeal; reroute to a sibling venue or journal, and mind the 12-month ban
A response moves borderline papers when it corrects a factual misreading or supplies a measurement a reviewer said was missing; it does not move papers when it argues taste. In a revise, silent omissions — a requested change neither made nor explicitly declined with a reason — are what turn the second read into a rejection.
Reading a review packet
Weight reviews before answering. A review that cites your figures, tables, and configuration was read closely and will be read closely again in the revision — its author is your likely advocate if the changes hold. A review that discusses only novelty has left soundness and reproducibility to the others; answer each reviewer on the axis they raised. Reviewers often end with an explicit request list; the revision is scored heavily on whether each request got a direct, evidenced answer.
Misreadings to avoid
- Treating a revise as a guaranteed accept — the in-cycle second read is real; budget the revision window like a deadline.
- Treating the author-feedback phase as the whole game — it is short and decision-focused; the revision carries the real argument.
- Assuming the last cycle presents this year — it rolls to the next edition.
- Confusing EDBT and ICDT reviewing — separate calls, separate PCs, despite the joint week.
- Projecting last year's cadence — cycle count and per-cycle timing are decided per edition.
Output format
[Process stage] pre-submission / author-feedback / revise / final / accepted
[Cycle] which cycle, presents at <edition/year>
[Decision category] accept / revise / reject, with the criterion driving it
[Criterion map] each review point -> significance | soundness | evaluation | reproducibility | clarity
[Leverage plan] the next-stage action that can actually change the outcome
[Forbidden moves] unsupported new claims / (if double-blind) identity leak