15

I was asked by Program Committee members to be a sub-reviewer for some papers related to my research. The PC members are big shots and I really want to impress them.

It is easy to write a review if I can discover weaknesses in the paper. However, if the paper is strong or I am unable to find any weaknesses, I feel my review is useless.

Unlike the flaws, the authors tend to repeat and highlight their contributions. So my review is just rephrasing the contributions and how I am impressed by them.

If I'm unable to give any suggestions to improve the approach or the narrative of the paper, what should I do to write a good review? How to show the PC members that I didn't just skim through the contributions, and agreed with the authors (strong papers are often written by well-known researchers)?

|improve this question
  • 2
    Would you mind expanding PC? Does it stand for paper committee? – mkennedy Jul 23 '15 at 16:43
  • @mkennedy: I am used to PC serving as an abbreviation for "program chair[persons]" and therefore implicitly assumed it meant that, but the author should indeed clarify. – O. R. Mapper Jul 23 '15 at 21:20
  • @O.R.Mapper that makes sense. I understood the question, but couldn't decide if the OP was reviewing for a conference or journal, or something else and thought expanding PC might help to clarify the point. – mkennedy Jul 23 '15 at 21:24
  • 2
    @mkennedy: PC stands for Program Committee, those who review or assign sub-reviewers in (CS) conferences. – qsp Jul 23 '15 at 21:29
  • 2
    While it is understandable to think in terms of "impressing", this is a very corrupting element to add to one's thinking. I'd recommend trying to pose and answer (for yourself, for example) the question without such self-emphasizing qualifications/words first, before starting to think about the professional game-playing aspect. Sure, the latter is present, and often very important, but complete prostitution is undesirable. – paul garrett Jul 23 '15 at 23:21
14

The key principle is this: just as with a negative review, you need to demonstrate that you have read the paper thoroughly and are making your judgement for sound scientific reasons.

If you like a paper, you should be able to say why you like it. Beyond the paraphrased summary of goals and contributions (which should begin any review), you can spend a sentence or two saying what aspects of the work or its presentation you found particularly compelling or particularly elegant, and why.

It's also a rare paper that doesn't have any minor errors or needs for improvement, and you can also point out those, prefaced with something like, "The only things I see needing improvement are a few minor points of copyediting and clarification."

In the end, your review might be quite short, and that's fine.

|improve this answer
  • 2
    I suppose you implicitly mean this, but it should probably be explicitly written down: If you can find a few minor issues this is a good indicator that you actually thouroughly read the paper. In particular if you are doing a regular review, it’s more beneficial to the editor and author if you list some minor issues than if you have nothing negative to remark at all. – Wrzlprmft♦ Jul 23 '15 at 15:45
  • 2
    @Wrzlprmft True. I have however, had a couple of occasions where I literally had nothing bad to say about the paper because the authors were apparently better proof-readers than I. Which is why I think that giving good details on why you like the paper is an even better way to show that you have read and comprehended it. – jakebeal Jul 23 '15 at 15:52
10

It is easy to write a review if I can discover weaknesses in the paper.

That is because you comment on the weaknesses. You do this to show to someone who is convinced of the paper why you think the paper is weak.

For a strong paper, you would do the opposite: You highlight the strong points, in order to show to someone who is convinced that the paper is poor that it is not.

If it helps you, you can create a checklist for yourself, listing aspects that you expect to find in the paper. For instance:

  • Is the contribution clear?
  • Do I understand all parts of the novel idea?
  • Are there any examples of how to use the novel concept?

Often, some of these questions are already provided in the invitation for the review. In any case, you can highlight the answers to these questions in your review, and thereby corroborate your suggestion to accept the paper.

In all the cases, I'm unable to give any suggestions to improve the approach or the narrative of the paper.

You do not need to do this. If you really think the paper is great, and if that impression is supported by the aforementioned checklist, there is no point in suggesting changes for the sake of providing some suggestions.

How to show the PC members that I didn't just skim through the contributions, and agreed with the authors

That is exactly one of the reasons why you should explicitly highlight which parts of the paper you liked. One way to show that you actually understood the content of the paper is by writing a short summary of the paper in your own words at the beginning of the review. Later on, when you highlight positive aspects of the paper during the further text of your review, you can refer to the various parts of the paper content that you already mentioned during your short summary. This should be sufficient to show to both the authors and the PC members that you based your positive evaluation on actual comprehension and thought, rather than a cursory impression.

|improve this answer
3

Just write a thoughtful review on why you think the paper is strong. If the paper is well written, then say how it is well organized, and why the idea is clear; again, justify, justify! You may include its significance and how it will contribute to the discipline as a whole. ...

You will give a good impression if you show that you have been diligent and reflective in your write-up.

|improve this answer

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.