Information for Authors

Right here. While the occasional update to the timeline might be unavoidable, we will try to always keep the information on this page up to date. Note that anything that qualifies as a deadline always comes into force at the end of the day specified, anywhere on Earth (UTC-12).

• Submission site opening: Monday, 1 April 2024

• Abstract deadline: Friday, 19 April 2024

• Paper deadline: Thursday, 25 April 2024

• Rebuttal period: Monday-Wednesday 17-19 June  2024 (72 hours)

• Author notification: Thursday, 4 July 2024

• Early registration deadline: Monday, 19 August 2024

• Camera-ready deadline: Monday, 26 August 2024

For important dates regarding PAIS, workshops, tutorials, the doctoral consortium, the demo track, or the journal track, please refer to the respective Calls on the conference website.

The Call for Papers for the main track is here: https://www.ecai2024.eu/calls/main-track

Note that the main track is not the only opportunity for submitting to ECAI. You can also propose a workshop (or submit to such a workshop), you can propose a tutorial, you can submit a demo, you can present in the journal track (if you have an eligible journal paper), you can apply to participate in the doctoral consortium,  or you can submit a paper to PAIS (our sister conference on applications of intelligent systems). Refer to the conference website for information on all of these opportunities.

Here it is EasyChair site for the main track: https://easychair.org/conferences/?conf=ecai24

For links to the EasyChair sites for PAIS, workshop proposals, tutorial proposals, the doctoral consortium, and the demo track, please refer to the respective Calls on the conference website. 

You can reach the two PC chairs by sending email to pc-chairs@ecai2024.eu or ecai24@easychair.org. This will reach both Ulle Endriss and Francisco Melo (and nobody else).

ECAI is a discipline-wide conference and welcomes submissions on all aspects of AI. When you submit, you will need to associate your paper with two or more 'topics' from a long list of topics.

Probably yes. The list of topics is an attempt to categorise most of the submissions we expect to receive, but of course no such list can ever be exhaustive. If, after having read a number of papers published in the proceedings of ECAI (or the closely related conferences IJCAI and AAAI) in recent years, you feel that your paper is a good fit for the conference, then please submit.

If you do not find a topic in the list that is an obvious fit for your paper, the best strategy is to pick topics that you think will attract the attention of the kind of people you would like to review your paper.

The Call for Papers explicitly forbids concurrent submission (also known as dual submission). So, if you have submitted or will submit a variant of your paper to another conference, workshop, or journal, then you cannot submit to ECAI in case the reviewing periods overlap. There are two reasons for this.

First, submitting to multiple events in parallel can be seen as an attempt to undermine the integrity of the reviewing process, by trying to exploit the element of randomness that necessarily is part of every such process.

Second, in the interest of producing good science and out of respect for the work done by reviewers, authors should always use the feedback received in one reviewing process to improve their paper before submitting it to the next reviewing process. Note that this argument applies also in case the other event is a workshop without archival proceedings.

For ECAI we review full papers of up to 7 pages (plus up to 1 page of references). Your paper should start with a short abstract of roughly 150-250 words. A good abstract will indicate the broader research area you are contributing to, outline the research question you are addressing or the problem you are trying to solve, sketch the results obtained, and mention the methods used. It should allow a potential reviewer to assess whether they want to read the full paper and would be qualified to evaluate it. 

You are required to submit a copy of this abstract (in plain text) 6 days before the main paper submission deadline. That’s the abstract submission deadline. We need this information to be able to assign reviewers to papers so as to be able to start the reviewing process in a timely manner. For this reason, you should not make significant changes to the abstract of your paper after the abstract submission deadline. Minor reformulations are fine. 

Yes, you can update your paper (and the supplementary material, if applicable) as many times as you like before the paper submission deadline. Your reviewers will only get to see the final version you submit.

Similarly, you can update the abstract of your paper as many times as you like before the abstract submission deadline.

No, this is not possible.

Please note that minor mistakes, such as not including the submission number on the paper itself (or keeping the #123 from the sample file rather than inserting your own submission number) or not enabling line numbering are not the end of the world. These things won't get your paper desk-rejected, and there is no need to email us.

We are aware of the fact that the ECAI LaTeX template does not work well with certain packages (such as subfig, subcaption, and hyperref ). Sorry about that. Hopefully, the number of authors affected by this is very small.

Please be pragmatic about this and find a workaround for your specific situation yourself (without touching the class file). If it renders correctly, then it is perfectly acceptable to us.

Specifically, you might succeed in working around some of the issues created by the subfig package by manually inserting positive vertical space yourself, using the \vspace-command. Alternatively, you could try the older subfigure package, which appears to work better in the context of the ECAI LaTeX template. 

Double-blind reviewing means that reviewers should not be aware of the identity of the authors of the papers they review (on top of authors not being aware of the identity of their reviewers). 

The idea is that double-blind reviewing will make it less likely that the work of well-known authors (or authors from well-known institutions) will be treated more favourably than everyone else's submissions. More generally, double-blind reviewing is intended to reduce the impact of the various unconscious biases most reviewers will be subject to. This not only makes the system fairer but it also improves the quality of the research we publish as a community.

So make sure your paper is fully anonymised. Obviously, at submission time, the names of the authors should not be shown below the title of the paper. Any acknowledgements should also be omitted. The ECAI LaTeX template takes care of these things automatically, if used correctly. Don't forget to double-check any parts of your paper where you talk about your own prior work.

Papers that clearly violate the anonymity requirement will be desk-rejected.

Yes, uploading your paper to arXiv or another preprint server before submission (or during the reviewing period) is permitted (though not particularly encouraged).

The reason why it is not particularly encouraged is that doing so, especially doing so around the time of the submission deadline, can undermine the double-blind reviewing policy of the conference, as it increases the chances that your reviewers will see an announcement of your preprint around the time they start working on their review. But if you feel that you have a good reason for putting your paper on arXiv at this particular time, there are no hard rules against doing so.

For the same reason, actively promoting your paper on social media channels during the reviewing period is also discouraged.

Yes, you can cite yourself. To decide whether doing so would be appropriate, conduct the following simple thought experiment. Imagine another researcher were to write the paper you are working on. Then, would an expert in your field say that this other researcher should cite your prior work? If yes (and only then), you should cite it as well.

Of course, to guarantee anonymity, you must use the third person rather than the first person when referring to your own prior work (thus, "X et al. [42] showed ..." rather than "We showed ... [42]"). If your paper gets accepted, then you may want to reformulate this again (it can sound a bit weird if you are referring to yourself in the third person in a published paper!).

Do not edit out the occurrences of your own name in the bibliography. First, that would only make it more obvious who you are. Second, that would make it harder for reviewers to consult the papers cited in the bibliography.

Provided the earlier work has not been published (or is due to be published) in an archival forum, this is ok.

If the paper you are submitting is a variant of a paper previously presented at an informal workshop, then do not cite that earlier paper. Doing so would make it impossible to maintain anonymity. The same applies in case a variant of your paper is available as a technical report or preprint, or in case your paper is based on your thesis. 

If your paper gets accepted, then you can cite those earlier versions of your work in the camera-ready version of your paper, and it is up to you to decide what would be appropriate in that case. 

Yes, you can submit such supplementary material together with your paper, and it is good practice to do so for any paper that reports on an experiment involving data or code. Put everything in a single ZIP file and upload it separately from your paper. Make sure that all of it is fully anonymised.

First of all, provided you do not go beyond 7 pages (plus 1 page of references), you of course are free to call one of the sections of your paper an "Appendix".

Then, yes, uploading a separate document as supplementary material is possible. There are no formal restrictions on length or style. Make sure everything is fully anonymised. If you are submitting supplementary material including more than one file (say, a document and some code), then put everything in a single ZIP container before uploading it.

You must not include your appendix in the paper itself if that would violate the page limit.

You must make sure that any supplementary material you submit really is supplementary in nature: any information that is essential for understanding or evaluating your paper must be included in the paper itself.

For example, it is not acceptable to relegate most of the proofs of your theoretical results to the supplementary material. If the core of your contribution is, say, a theorem that requires a 10-page proof, then ECAI is not the right venue for your paper. In such a case, you should submit to a journal instead. (If you submit to AIJ or JAIR, you might then present your work at the journal track of a future edition of ECAI.)

When working on a paper for a conference, it is common that your first complete draft is way too long, given the page limit imposed by the conference organisers. These limits are there, first and foremost, to protect the time of the reviewers (and in the old days to save paper), but in our experience, being forced to revise (and revise again and again) a paper to make it fit can also greatly improve the quality of your writing. Explanations have to be formulated more carefully, easier proofs have to be found, tables with less exciting experimental results have to be omitted or replaced by more insightful prose, and so forth. So please do not interpret the availability of the supplementary-material option as a license to circumvent this crucial stage in the research process and simply cut off the last few pages just before the deadline and declare them "supplementary material". This almost certainly will result in a poorly presented submission that will have little chance of getting accepted.

We recommend that you work under the assumption that your reviewers will not actually look at the supplementary material you submit.

No, it does not mean that. Including a proof sketch in a full paper can be perfectly reasonable. If you are able to provide enough information so that an expert would be able to fill in the missing details and verify the correctness of your claimed result, then this is absolutely fine. Indeed, sometimes leaving out certain details can greatly improve readability without really affecting reproducibility of the results. 

For example, for an NP-hardness proof it will sometimes make sense to just indicate the reduction used but to leave it to the reader to verify the correctness of that reduction. Another situation in which a proof sketch is perfectly acceptable is one where you are using the same technique several times in a row to obtain similar results. If you illustrate your approach properly in the context of the first such proof, then it will often be perfectly acceptable to reduce the remaining proofs to very short sketches that only hint at the main difference relative to the first proof.

No. This would violate the integrity of the reviewing process, even if you really manage to fully anonymise the material. Crucially, we do not have any means of ensuring that the material will remain unaltered throughout the reviewing period. Also, it is conceivable that an author might try to track who is accessing the material, thereby violating the anonymity of reviewers.

Authors who violate this rule, and include such a link in their paper, risk getting their paper desk-rejected.

We note that we have seen several cases where repositories hosted on supposedly "anonymous" sites (such as https://anonymous.4open.science/) in fact are not anonymous at all. For example, some include the name of an author in a copyright notice and others include links to arXiv reports by the authors. 

We also have seen cases where authors indeed have changed the material provided during the reviewing period (in particular, we have seen cases where they have removed their repositories altogether). 

The basic principle at work here is that the readers of your published paper should have access to the same information as the reviewers of your submission.

So you should make your (de-anonymised and suitably revised) supplementary material openly available in archival form at the time of publication of your paper. For code or data you may wish to use a service such as Zenodo. For a technical appendix with additional proof details or similar, you may wish to use an archival preprint server such as arXiv. Try to find a place for your supplementary material where you can be reasonably sure that it will still be available, say, 50 years from today. 

It is good practice to include a reference to the supplementary material in the camera-ready version of your paper (listed as an item in your bibliography). It should be possible for people to cite your supplementary material independently from citing your paper.

Remember that a simple reference such as "see appendix" does not make sense in your published paper, as the appendix won't be published with your paper. 

Be brief. Be polite. Try to answer all concrete questions raised in the reviews. Try to make it very clear which point you are responding to (this should be clear also for, say, an area chair or a SPC member who did not read your paper in any depth).

Avoid the temptation to argue with your reviewers about matters that are subjective in nature (such as the importance of a result). You are unlikely to convince them with just a short paragraph of text. But do try to correct factual misunderstandings of your contribution.

If you can indicate a couple of very concrete changes you plan to make to your paper to address the concerns of reviewers, then do. But avoid making vague or far-reaching promises. Keep in mind that we need to evaluate the paper you submitted, not the paper you promise to write in the future.

Of course, make sure you do not compromise the anonymity of your submission. 

We have opted for rebuttals in plain text to keep things simple, both for you as authors and for the PC members reading the rebuttals. You are not expected to spend a significant amount of time on this, and the simple format chosen reflects this.

Keep in mind that the purpose of the rebuttal is to respond to simple clarification questions, not to report on additional research you are trying to conduct during the rebuttal phase itself.

Never include a link to additional supplementary material in your rebuttal. This is part of the policy regarding such links described in the Call for Papers, which is explained in detail elsewhere on this page. 

We will desk-reject submissions that violate this rule.

If a reviewer is unaware of this rule and asks for a link (say, to a codebase or additional experimental data), please simply point out that you cannot comply due to this rule. Of course, if what they actually are saying is that you should have submitted your code as supplementary material, then they may very well have a point.

Naturally, including a link to publicly available information is fine (say, to a paper disproving a factually wrong claim by a reviewer). In case this is one of your own publications, do not reveal your identity .

Producing a revised version of your paper is not part of the rebuttal process.

During rebuttal, you only answer to the questions received and indicate how you plan to revise your paper in case it should get accepted. The idea is to make the workload manageable for both you and your reviewers.

If you paper gets accepted, you will then be expected to revise your paper before it gets published. This is what you do between the notification date and the camera-ready deadline mentioned in the Call for Papers. You are expected to make a best-effort attempt to incorporate the feedback obtained and, of course, to fix any mistakes you noticed yourself in the meantime. Naturally, if you have good reasons for not taking on board a specific piece of advice, this is perfectly acceptable. Note that making very substantial changes that would change the essence of the paper, such as adding new results that have not been reviewed, is not permitted.

If a reviewer raises doubts about whether your work is sufficiently original for publication relative to some other paper in the literature and that other paper is actually one of your own papers, then it can be tricky to respond to this without breaking anonymity.

If the other paper is a formally published paper reporting different (but related) results, then you should respond to it in the same way as you would if the other paper were somebody else's work. Once your prior work has become part of the literature, your new work needs to distinguish itself from that prior work in the same way as it needs to distinguish itself from prior work by others.

If the other paper is an earlier version of your submitted paper that you presented at a workshop (without formal proceedings) or that you made available as a preprint (e.g., on arXiv), then please send us (the PC chairs) an email to notify us of the misunderstanding. We will then be able to clear up the misunderstanding in a manner that preserves anonymity as best as possible. The same applies in case the overlap in contribution noted by a reviewer concerns work in a Master's or PhD thesis of yours.

No.

We were aiming at providing 3 reviews for every submission (that did not get desk-rejected on formal grounds). Every single paper indeed received at least 3 reviews, but a small number of papers (around 10%) got more than that. There can be different reasons for why a paper receives more than 3 reviews, and if yours did you should not interpret this as either a bad or a good sign.

In most cases, one of the original reviewers was late and in the end several of the individuals we approached for replacement reviews delivered (and sometimes the original reviewer still delivered as well).

In some cases, the views expressed in the first 3 reviews differed more than what one would usually expect to see, so it seemed helpful to ask for an additional opinion. In some cases, one of the original reviews was very short or the reviewer expressed a low level of confidence in their review. Of course, there are also many papers that fall into this category where we did not solicit (or receive) an additional review.

We really hope you won't, and we have put mechanisms in place that we believe will make it unlikely that you will (such as carefully selecting PC members, providing clear instructions for less experienced reviewers on this site, giving them sufficient time to write their reviews, and asking SPC members to read all reviews before you receive them). But no system is entirely foolproof and most of us will receive a couple of unprofessional reviews at some point in their career.

These are typical signs for unprofessional reviews:

• The review includes rude or otherwise inappropriate language.

• The review is very short (just a few lines of text), it only makes generic statements (rather than focusing on the specifics of your paper), or it shows clear signs of having been computer-generated.

• The review does not focus on the content of your paper but immediately drifts off into only vaguely related territory.

Regarding the first point, if you have served a few times as a member of a programme committee yourself, you will know from experience that authors sometimes interpret colloquial language in a review quite differently from how it was intended by the reviewer. So please always try to give the reviews you receive the most charitable reading you can. It really is extremely rare that a reviewer actually wants to insult you.

Regarding the second point, please note that SPC members and area chairs are instructed to give little weight to reviews that are clearly superficial (both the positive and the negative ones). So, while it is regrettable when you receive such a review, there may be no need to submit a complaint about it. 

Please note that the following cases do not qualify as unprofessional reviews:

• The reviewer does not seem to grasp parts of your paper that you consider very basic.

• The reviewers asks a question already answered in the paper or laments the omission of information that in fact is there already.

• The reviewer makes a suggestion for a technical improvement that you already know to be wrong or unfruitful.

• The reviewer admits to not being an expert in your area of research.

• The reviewer is unaware of the rule saying that you cannot provide links to code (but that you should instead have uploaded code as supplementary material). Please simply remind them of the rule.

• The review has grammatical mistakes or looks messy. 

• The reviewer disagrees with you about the significance of one of your findings.

• The reviewer reports a numerical score that you consider too low.

• The reviewer seems to not have read your supplementary material. Please recall that they are not obliged to.

• The reviewer seems to not have noticed the resubmission information your provided. Please simply remind them that you provided this information.

• The reviewer shares a negative opinion about the timeliness or importance of the entire research area you are contributing to. They might feel they are giving good advice, so please take it in good faith. We won't reject papers on such grounds.

Please also note that (of course!) different reviewers will sometimes disagree amongst themselves about the merits of your paper. This also is not a sign of lack of professionalism.

So, what do you do if you receive a review that really is unprofessional? In most cases, the best thing to do will be to politely address the issue in your rebuttal. The other reviewers and both the SPC member and the area chair assigned to your submission will get to see all reviews and your rebuttal. If you highlight a specific problem with one of the reviews, then they will get to discuss that point and, if appropriate, disregard the issue when making the final decision on your paper. The general advice for writing rebuttals applies also here: make it very easy for people to understand what you are referring to.

If you feel the situation is too delicate to address in your rebuttal, then you are welcome to send us (the PC chairs) an email. Before you do this, discuss the situation with all your coauthors (if any) and recall that in such a situation it is always a good idea to take time to reflect and wait for 24 hours before sending off your email. Keep your email short and explain your concern in a way that allows us to assess the situation without reading your paper (but only the offending review). 

We will do our best to respond in a timely manner, but please do not expect an immediate reply. We will only process emails that (1) mention the submission number in the subject line, that (2) start with a statement saying that you have read the advice provided here, and that (3) are signed by all authors (who all must be in cc). 

If you wish to withdraw your paper, please send an email to pc-chairs@ecai2024.eu with the words "withdrawal request" and the submission number of your paper in the subject line. The message should be signed by all authors (and they must all be in cc).

We will accept all withdrawal requests that reach us by the end of the rebuttal phase, but it may take us up to one week to process your request. 

Whether or not a withdrawal request received later than that will be accepted depends on how far advanced the decision process on your paper is at that point. Basically, we will accept it if (and only if) doing so will save the programme committee some work.

One you have received our confirmation that we accept your withdrawal request, you are free to submit your work to another conference (without violating the concurrent submission policy).

No, this is not possible.

By the time of the start of the rebuttal phase, we had desk-rejected some 7.5% of all submissions. (Yes, that's a lot!) This includes fewer than a handful of papers rejected for being out of scope or not meeting basic standards of scholarly publication. All other desk-rejected papers were rejected on purely formal grounds, for violating one of the policies stated in the Call for Papers. By their very nature, these are objective grounds for rejection, which is why we cannot enter into any discussion about these cases.

Most of these were clearly honest and unfortunate mistakes by the authors, such as accidentally revealing their identity in the paper or including a link to online supplementary material rather than uploading that material together with their paper. If this applies to you, it should be easy to avoid the same mistake for your next submission. We wish you the best of luck for that next submission!

But a significant number of desk-rejections are also due to attempted fraud, most notably the concurrent submission of the same paper to multiple conferences. For the most egregious cases we are looking into additional sanctions for the authors concerned, besides just rejecting their papers from all of these conferences.

No, this won't be possible.

ECAI wants to be a meeting place for the international AI community, facilitating scientific networking and a lively exchange of ideas. This works best when people meet in person.

We of course have considered the possibility of running the conference in hybrid mode, with some presenting in person and some online. But this would significantly lower the quality of the experience for those present on site: Travelling to a conference to then sit in a (possibly almost empty) room to attend a streamed or pre-recorded talk can be a very frustrating experience. And it would not be satisfactory for the remote participants either, as their contributions would get very little attention from others. Finally, running a conference in hybrid mode would significantly increase costs, which in turn would require noticeable increases in the registration fees for everyone.

See also the attendance policy on the ECAI website.

Every paper accepted for the main track will get assigned either an oral presentation slot (talk) or a combined poster/spotlight presentation slot. No paper will get both a full talk and a poster. The presentation modality is not intended to reflect scientific quality. In particular, no distinction will be made in how papers are listed in the proceedings.

To make it clear to everyone that both groups of papers have the same status, the decision of which papers get presented by means of talks and which ones get presented by means of posters will be made in a randomised fashion, subject only to basic constraints intended to ensure that we get thematically coherent technical sessions. This will give everyone the same chance to present orally and should reduce the impact of implicit biases the PC chairs might be subject to.

As we expect the oral presentation option to be more popular with most authors, if you specifically want to present by means of a poster, this will be possible.

Oral presentations will get 15-minute slots, spotlight presentations will get 2-minute slots, and poster sessions most likely will last 90 minutes.

Yes. Here's a screenshot.

Note that at the end of the rebuttal phase we added an additional (optional) field intended specifically for comments on author responses to the initial reviews.

No, this is not possible. All decisions of the programme committee are final.

We fully acknowledge that such editorial decisions are never perfect. People make mistakes. And people can disagree on what would be a reasonable standard of acceptance. In particular, for a conference such as ECAI there are always a large number of borderline cases where the decisions conceivably could have gone either way.

Let us briefly react here to a few common concerns that authors who are pondering an appeal might have.

First, we stress that scores alone do not tell the full story. So there is no magic number x such that we would automatically accept all papers with an average (or a minimum) score of x.

Second, we are aware of the fact that not all reviewers followed our instruction to always react to an author rebuttal, at least if that rebuttal includes an attempt to answer to one of the concrete questions posed by said reviewer. This shows a certain lack of professionalism on the part of the reviewer and can be very frustrating for the authors. But we unfortunately cannot do anything about this. In particular, we will not attempt to chase reviewers for those missing reactions still after the notification date (we already did a lot of chasing before decisions were announced).

Third, for any serious problems with individual reviews, there was an opportunity to flag those during the rebuttal phase, so there should be no need to discuss those after the notification date. Of course, for a small number of papers an additional review was written after the rebuttal phase. Unfortunately, it is not feasible to organise a second rebuttal phase for those extra reviews.

If we still receive an email with an appeal against a decision by the programme committee, we will always check that no procedural mistakes were made and that the final decision communicated to the authors really reflects the intent of the programme committee. But we will never engage with arguments put forward by authors that relate to the content of their paper. Doing so would not only be completely infeasible for us, but it also would be disrespectful to the PC members who read and evaluated the paper in much more depth than we ever could.

Hard no.

Seriously: Please just write to us in your own words, even if not in perfect English, and keep your messages short and to the point. Those long-winded and obviously computer-generated "letters" we have been receiving will only slow us down and won't make it any more likely that we will accept a request we otherwise wouldn't.