38

I sent a paper to a top-tier computer-science conference, it was rejected and I got several useful suggestions for improvement. I improved the paper accordingly and I consider to submit it to the same conference in the following year. It is likely that the paper will be reviewed by the same referees.

When sending a revised paper to a journal, the authors should indicate where exactly they revised the paper, so that the referees do not have to read everything again. Should I do the same in this case? If so, how can I indicate to the referees the locations in which I corrected the paper according to their comments from previous year?

My concern is that, if I do not indicate where I revised the paper, the referees might think "we have already reviewed this paper and rejected it last year, and it looks similar, so we should reject it again".

  • 7
    I have upvoted your question as I believe it is worth addressing. However, there is a reason that conferences have become the primary venue by which CS disciplines publish their work: because it's faster than publishing in a journal! So, my question to you is: why not find another top tier conference? In my sub-discipline their are 4 or 5 to which I can submit (to be fair, not all of 100% equivalent prestige, but still top tier...) at different times throughout the year... – DMML May 20 '16 at 14:17
  • 2
    @DMlash is right. There is no need to send your papers to only one conference. I also do not agree 100% with "It is likely that the paper will be reviewed by the same referees". Every conference has different PC chairs and possibly many of its members every year. Why do you think you will fall into the exact same 3 reviewers? – Alexandros May 20 '16 at 14:49
  • 4
    I also do not agree 100% with "It is likely that the paper will be reviewed by the same referees". — This really depends on the paper, and perhaps on the subfield. For many conferences, most initial reviews are not written by PC members, but delegated by PC members to external experts. For very specialized papers, the set of experts can be small enough that repeated reviews by the same people are likely. I can think of several papers that I reviewed three or four times before they were accepted or the authors gave up. (And yes, if they ignored my first review, by 2nd review was short.) – JeffE May 20 '16 at 16:13
  • 1
    @DMlash That depends strongly on the research area. In some areas, the number of top-tier conferences is small, and they are not evenly distributed over the year. (For me, all the "big" conferences are in Summer.) – Uwe May 20 '16 at 16:19
  • 2
    @Alexandros Sure, if there are just a handful of potential reviewers, they will get the paper, no matter what conference it's sent to. – Uwe May 20 '16 at 20:47
33

I would not recommend stating that it's a revised paper for several reasons:

  1. Contrary to your assumption, it is not likely to be reviewed by the same people. Typically, each topic in a large computer science conference may be covered by many reviewers, and they get assigned either arbitrarily or by a bidding process. With a few reviewers assigned from a large pool, there might or might not be overlap, but the majority will likely not have seen the prior version.

  2. Computer science conferences, unlike journals, do not tend to provide a "comments to reviewers" section in their submissions. Trying to wedge that in will likely make your submission stand out in a bad way, putting the idea of rejection in your reviewers' heads.

  3. If you've really done a strong set of revisions, it should be pretty obvious that the paper has changed a lot even to a reviewer who's seen it before.

In short: I see no particular positives and several negatives for such a declaration.

3

Having to review the same paper a second time happened to me personally. we are a small phil. journal and thus have (so far) only a very limited pool of reviewers.

On one hand the submitter might have decisively improved his paper and thus it absolutely deserves to be treated like any other submission. Anything else would be unfair.

On the other hand I also have to be aware that also I, as a reviewer, can make mistakes. So even if it is precisely the same paper I read last time, I work as diligently as if I saw it the first time. What if it was my misunderstanding that made me reject it the first time?

Long story short: Every paper submitted has to be treated the same, anything else would be very unprofessional for a reviewer.

Include the reviewers' suggestions but be daring enough to also discuss them. After all, also I, as a reviewer, am but a human being and can be wrong. The one thing I love most about reviewing papers:

I, the reviewer, can learn a lot from the submitters and get the chance to rethink my own position!

2

In some CS areas the conferences are very competitive, so it's a little surprising that just by addressing previous comments, the paper will be good enough for next year. It would need to show it's better than the literature published this past year as well.

That said, if this is a largely experimental field, then it's reasonable to assume that if you weren't better than, say, the state-of-the-art before, and you are now, you will be accepted even if it's a revised paper. After all, in journals they ask whether or not the manuscript has been rejected before, and in conferences they do not. If they wanted to know - they would ask.

As a reviewer, even if I might recognize the name from last year's conference, I would check it again thoroughly and possibly try to see where you improved. So as the author, I would try to do my best to ensure that my paper has all minor issues fixed. If the same mistakes are repeated, then in the off-chance you will get the same reviewer, you might get an ill favored response - even more so than if it had been an original paper.

  • "so it's a little surprising that just by addressing previous comments, the paper will be good enough for next year" - being competitive means that submissions have to be more convincing than many other submissions to the same conference, and the standards for being considered as "convincing" are high. This does not automatically mean that every given topic or question receives attention in submissions in each year, and thus it is fully possible to improve a paper (e.g. by conducting a more significant user study with 50 participants instead of the original 5) while no new literature has ... – O. R. Mapper May 21 '16 at 17:53
  • ... been published on the specific question investigated in the paper. – O. R. Mapper May 21 '16 at 17:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.