ase-topic-selection

ASE Topic Selection

Decide the venue before drafting. ASE — the **IEEE/ACM International Conference on Automated

Software Engineering — is the home for research that automates** software-engineering tasks:

the analysis, design, implementation, testing, comprehension, and maintenance of software, and

increasingly the two-way street of AI for SE and SE for AI. Its reviewers read for a **technique or

tool that automates something**, evaluated on real subjects. A paper whose real center is a broad

empirical observation, a testing-theory result, or a pure ML advance is respected and then rejected

as off-center.

The routing question that matters most

The decisive ASE question is not "is this SE?" but **"is the contribution an automation — a

technique or tool that does an SE task a human or a weaker tool did before?"** If the beating heart

of the paper is a new automated mechanism (an analysis, a generator, a synthesizer, a repair

technique, a learned model applied to an SE task), ASE is a natural home. If the heart is a finding,

a theory, or a measurement study, a sibling fits better — even when the topic overlaps.

Sibling-venue routing table

| Signal in your project | Better home | Why |

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

| Core is a technique/tool that automates an SE task, evaluated on real systems | ASE | Automated software engineering is its named center; dual IEEE/ACM, IEEE-Xplore+ACM-DL proceedings |

| Broad empirical or foundational SE contribution; the lesson is a finding, not a tool | ESEC/FSE | The general-SE flagship; PACMSE journal model, empirical breadth |

| Equally broad SE technique, but ICSE's cycle/stage fits and you want its reach | ICSE | The other general-SE flagship; different calendar and (IEEE-side) format |

| Depth in testing theory, program analysis foundations, or fault detection | ISSTA | Deeper testing/analysis expertise, round-based reviewing |

| The automated mechanism is a program-language/compiler advance | PLDI / OOPSLA / POPL | PL venues own the language-mechanism contribution |

| The contribution is a new model/architecture; the SE task is a benchmark | an ML venue | Model-as-contribution routes out of SE |

| Study too long or too nuanced for the page budget | TSE / TOSEM / EMSE | Journals with no page ceiling; some feed ASE's Journal-First track |

Contribution shapes ASE rewards

• Technique + tool + evaluation on real subjects — a new analysis, test-generation, synthesis,

repair, or comprehension technique, embodied in a tool and run on real systems against credible

tool baselines (the TestEra / runtime-monitoring lineage in the exemplars library).

• Automated technique with a disciplined empirical evaluation — pairing an automatic method with

an honest comparison and effectiveness metric (the Tarantula fault-localization lineage).

• Learning-based automation with the SE task as the point — learned code representations or

models applied to a real SE task, where the task result, not the architecture, is the

contribution (the code-clone-detection lineage).

• AI4SE / SE4AI — using AI to automate SE tasks, or applying SE automation to build, test, and

maintain AI/ML systems; a growing ASE center of gravity.

• Tooling infrastructure and datasets — often routed to the co-located Tools and Datasets

track when the contribution is a usable artifact more than a research claim.

The model-swap and re-label tests

Two quick tests sharpen a borderline verdict:

• Model-swap test: if your paper leans on an LLM or learner, ask whether the automation still

stands as a contribution when the model is swapped. If the design (the analysis, the coupling, the

oracle) survives, it is ASE-shaped; if only the model result remains, route to an ML venue.

• Re-label test: could this be submitted to FSE unchanged and read as a broad empirical SE

paper, or to ISSTA as a testing-depth paper? If its heart is automation machinery, ASE fits; if

its heart is a finding or testing theory, route accordingly.

Which co-located track fits

• Research Papers — a full technique/tool contribution with real-subject evaluation.
• NIER (New Ideas and Emerging Results) — a bold new direction not yet backed by a full

evaluation (4 pages): groundbreaking, reflection, or late-breaking-advances.

• Tools and Datasets — a usable tool or dataset as the contribution (short, demo-oriented).
• Journal-First — work already accepted at TSE/TOSEM/EMSE (after 1 Jan 2025) presented at ASE.
• Industry Showcase / Doctoral Symposium — practitioner results and PhD-stage work.

Cheap reconnaissance before committing

[Scope]   scan the last two ASE programs (dblp conf/kbse, conf.researchr.org) for your subarea
          -> 3+ recent papers = a reviewer pool exists; 0 = opening or mismatch
[Shape]   is your contribution a runnable automation, or a finding/theory? -> automation => ASE
[Citations] majority SE-automation venues (ASE/ICSE/FSE/ISSTA)? -> majority non-SE => naturalize first
[Calendar] compare the next ASE deadline with ICSE/FSE/ISSTA/journal dates -> nearest honest fit

Decision procedure

[Audience] who runs your automation if the claim holds? -> SE developers/testers/maintainers/tool-builders?
[Claim type] technique+tool / automated-technique+evaluation / learning-based automation / AI4SE-SE4AI
[Automation test] is a runnable mechanism the contribution? -> yes => ASE; no => FSE/ISSTA/PL/ML
[Track check] full technique -> Research; bold early idea -> NIER; usable tool -> Tools & Datasets
[Verdict] ASE research track / co-located track / sibling venue / journal, with a one-line reason

Run this before the writing skills; a wrong venue decision wastes every later step. When the

verdict is ASE, continue with ase-workflow for the calendar and ase-writing-style for the paper

shape.