技能 数据科学 数据库研究投稿最佳指南

数据库研究投稿最佳指南

v20260724
pods-topic-selection
本指南旨在帮助研究人员确定论文最合适的投稿平台,涵盖了从数据库理论到系统实现的投稿路径选择。它帮助区分论文是侧重于“可证明的理论定理”还是“实测性能的系统成果”,从而确保研究范围的准确性。
获取技能
429 次下载
概览

PODS Topic Selection

Decide the venue before drafting. PODS — the ACM SIGMOD/SIGACT Symposium on Principles of Database Systems — is the theoretical-foundations symposium held jointly with SIGMOD each year. Its papers are theorems about data management: models, query languages, complexity bounds, dichotomies, logic, and provably optimal algorithms. A technically strong paper whose real contribution is a faster system, a benchmark win, or an engineering artifact is respected and then rejected as out of scope — that paper is SIGMOD/VLDB/ICDE. PODS reviewers read for a provable, foundational statement about a cleanly defined model.

The routing question that matters most

The decisive question is rarely "is this about databases?" but "is the contribution a theorem or a system?" If the headline is a bound, a dichotomy, a semantics, an expressiveness result, or an algorithm with a matching lower bound, it is PODS-shaped. If the headline is throughput, latency, or accuracy on real workloads — even with clever theory inside — its home is a systems-DB flagship. PODS and SIGMOD share a week and a hallway but not a bar for acceptance.

Sibling-venue routing table

Signal in your project Better home Why
A bound, dichotomy, semantics, or provably optimal algorithm in a data model PODS The database-theory symposium; results are theorems with proofs
A system, index, or optimizer evaluated on real workloads for performance SIGMOD / VLDB / ICDE The systems-DB flagships; evidence is measured, not proved
Database theory, but you want the EDBT/ICDT federation or missed the PODS cycle ICDT PODS's sister theory venue; overlapping community and reviewer pool, different calendar
Pure logic / finite model theory with no data-management payoff LICS / ICALP / STOC / FOCS Theory venues; PODS wants the data-management motivation to be central
A deep, long development beyond a 15-page symposium result TODS / LMCS / JACM / VLDBJ Journals with no symposium page ceiling; often the full-version home
Applied ML-for-data with empirical validation as the point SIGMOD / VLDB / a ML venue Not a PODS theorem; route to where the evidence is measured

Contribution shapes PODS rewards

  • A new semantics or framework for an ill-defined problem — consistent query answering, provenance semirings, a data model for uncertainty — stated precisely and studied for decidability/complexity.
  • A complexity classification — a dichotomy (PTIME vs. hard), a fine-grained bound, a data-vs-combined-complexity separation over a query class.
  • An expressiveness or logic result — which logical language captures an operation (composition, view definition, path querying), with matching upper and lower bounds.
  • A provably optimal algorithm — a worst-case-optimal join, a constant-delay enumeration procedure, an MPC-round bound — the guarantee is a theorem, not a measured speedup.
  • Foundations of emerging data problems — differential privacy for queries, learning-theoretic guarantees for data tasks, graph-query theory — where PODS supplies the rigor.

The result-swap and re-label tests

Two quick tests sharpen a borderline verdict:

  • Result-swap test: if you replaced your specific algorithm or construction with a different one, would a theorem still remain (a bound, a classification, an impossibility)? If not, the artifact is the contribution and a systems venue fits better.
  • Re-label test: could this paper be submitted to SIGMOD unchanged and read as native there? If its heart is a measured system with theory as garnish, route to SIGMOD/VLDB; if its heart is a proof, PODS is home. The mirror also holds for ICDT — if the paper is pure logic with no data-management question driving it, ICDT or LICS may fit better than PODS.

Maturity, without the ladder cliché

Fit is necessary but not sufficient. A conjecture with partial evidence but no proof is a workshop or a Gems-of-PODS talk, not a research paper; a theorem whose proof only handles a special case needs the general result or an honest scope; a sprawling development that cannot breathe in 15 pages plus appendix may belong in a journal first, with a PODS extended abstract of the core theorem. Submitting a half-proved result costs a full cycle even when the topic is perfect.

Cheap reconnaissance before committing

[Scope]     scan the last two PODS proceedings (dblp db/conf/pods, PACMMOD PODS track) for your
            subarea -> 3+ recent papers = a reviewer pool exists; 0 = mismatch or ICDT/LICS territory
[Citations] is your bibliography majority PODS/ICDT/LICS/TODS/JACM, or majority SIGMOD/VLDB?
            -> majority systems venues => reviewers may read you as a systems paper; reframe the intro
[Calendar]  PODS runs two cycles a year; compare the next PODS cycle with ICDT's and the systems
            deadlines, and route to the nearest honest fit rather than idling

Decision procedure

[Contribution] theorem (bound/dichotomy/semantics/optimal algorithm) or measured system?
[Model]        is there a cleanly defined data/query model the result is about?
[PODS vs ICDT] data-management theory with the SIGMOD community pull -> PODS; EDBT/ICDT federation
               fit or calendar -> ICDT
[Systems check] performance is the headline -> SIGMOD/VLDB/ICDE, not PODS
[Verdict]      PODS / ICDT / systems flagship / journal-first, with a one-line reason

Run this before the writing skills; a wrong venue decision wastes every later step. When the verdict is PODS, continue with pods-workflow for the two-cycle calendar and pods-writing-style for the paper shape.

信息
Category 数据科学
Name pods-topic-selection
版本 v20260724
大小 6.03KB
更新时间 2026-07-29
语言