The CS Conference Submission Glossary

Every call for papers assumes you already know the vocabulary. Nobody teaches it, most of it is never written down in one place, and the terms that cost people the most are the boring procedural ones rather than the interesting ones.

Here is the vocabulary, with the actual timings from the 167 conference editions where we hold both a submission deadline and a notification date.

Before you submit

Call for Papers (CFP) The conference’s announcement of what it wants and by when. Read the whole thing, not the dates: the scope paragraph is what a desk reject will be measured against.

Abstract deadline / abstract registration Many venues make you register a title and abstract days or weeks before the paper itself is due. Miss it and you cannot submit the paper at all, even though the paper deadline has not passed. This catches people every single cycle. When a CFP lists two dates, assume the earlier one is a hard gate rather than a suggestion.

Some conferences invert this. The IATED conferences, for instance, are abstract-first: the abstract is the submission, and the full paper comes later if accepted.

Page limit, excluding references Almost always excluding references and appendices, but read it. Over-length is one of the most common desk rejects and one of the most avoidable.

Double-blind vs single-blind Double-blind means reviewers do not see your identity and you do not see theirs. Single-blind means reviewers know who you are. Under double-blind you must anonymise: no author names, no institution, no “in our previous work [12] we showed”, no GitHub links that resolve to your lab. Most CS venues are double-blind now; a few, including several IEEE conferences, are not.

Anonymity rules usually permit arXiv preprints, but not all of them do, and the ones that restrict it care. Check before you post.

While it is under review

Desk reject Rejected by the chairs before it reaches reviewers. Causes are almost always procedural: out of scope, over the page limit, broken anonymity, wrong template, or duplicate submission. It is the cheapest rejection to avoid and the most annoying to receive.

Rebuttal / author response A short window, typically a week, in which you answer the reviews before the final decision. Its purpose is narrower than people assume: it exists to correct factual errors and answer specific questions, not to re-argue the paper’s importance. Reviewers rarely change a score because you disagreed with them, and often do change one because you cleared up a misunderstanding they flagged.

Shepherding A conditional acceptance where a named reviewer works with you to get the paper over the line. If you are offered this, take it and do exactly what the shepherd asks.

Major revision / revise and resubmit Increasingly common in venues with multiple rounds: not accepted, not rejected, resubmit to a later cycle with the reviews carried over. VLDB and ICDE both work this way. Note that some venues explicitly do not offer it: ICDE’s research track is straight accept or reject.

How long this takes. Across our data the median gap from deadline to notification is 84 days, almost exactly twelve weeks. The spread is wide: the fastest we hold is 11 days and the slowest 153. Twelve weeks is the number to plan around.

After the decision

Notification The accept or reject. Note the date, because the next one comes fast.

Camera-ready The final formatted version, incorporating reviewer requirements, due typically two to four weeks after notification. People consistently under-plan this. If your notification is in mid-December, your camera-ready is due over the holidays.

Registration requirement / no-show policy Almost every venue requires at least one author to register and present, and many will pull the paper from the proceedings if nobody shows. This is why an acceptance is a travel commitment, not just a line on a CV.

The gap between yes and go. Median notification to conference across our data is 100 days, about fourteen weeks. That sounds comfortable. The tail is not: Eurocrypt 2026 notified authors six days before the conference opened. SIGCOMM 2025 gave 33 days. If you are waiting for the accept email before booking flights, check your specific venue’s gap first, because for some of them that plan does not work. We measured this properly in how conference timing actually works.

Submission models worth knowing

Multiple rounds or cycles A growing number of venues take submissions two or more times a year rather than once. This matters enormously for planning: missing a deadline becomes a few months rather than a year. Fifteen of the series we track run multiple rounds or cycles, including VLDB, ICDE, PETS, NDSS, ASPLOS, SIGMETRICS, UbiComp, AsiaCCS, CoNEXT, ICALP, HOST, ICICS and ACISP.

If a paper is three weeks from ready and the deadline is tomorrow, a second-round venue is usually the better call than a rushed submission.

Rolling review (ARR) ACL and EACL use ACL Rolling Review, where you submit to a central reviewing pool on a monthly cycle and then commit the reviewed paper to a specific conference. The consequence is that the deadline you care about is the ARR cycle date, not the conference’s own date, and the two are listed separately.

Artifact evaluation A separate, optional process after acceptance where a committee checks that your code and data reproduce your results, and awards badges. It is voluntary at most venues, increasingly expected in systems and software engineering, and worth doing: it is the cheapest credibility you will ever buy.

Terms that describe the venue, not the paper

CORE rank (A*, A, B, C) The Australian CORE ranking, the most-used tiering of CS venues. We explain what it does and does not mean in what are CORE rankings, and list every A* conference with 2026 to 2027 dates in the A* calendar.

Acceptance rate The share of submissions accepted. Useful, but far less comparable across subfields than people treat it: A* venues in our data span 14.0% to 43.0%. See the acceptance-rate analysis.

Main track, workshops, and co-located events The main track is the peer-reviewed conference proper. Workshops are smaller, usually same-week, often with their own lighter review, and are a legitimate first venue rather than a consolation prize. Co-located means two separate conferences sharing a venue and week.

Proceedings, LNCS, and indexing Where the paper ends up. Springer LNCS, ACM Digital Library and IEEE Xplore are the common ones in CS, and all are indexed. A workshop with published proceedings counts; one without does not, and CFPs are not always clear about which you are dealing with.

One practical note

A surprising amount of this is simply not published in advance. Of the 125 upcoming editions where we hold a submission deadline, 41 have no notification date announced at all. That is not sloppiness on the organisers’ part so much as a norm: many venues publish the date they need from you and not the date you need from them.

Plan on the medians, and check the specific venue when it matters. Everything we hold is on the deadlines page, sorted by what closes soonest.