I have found that one of the most common remarks from the reviewers is regarding the "state-of-the-art". Especially in terms of applied research. Something like:

  1. The authors have not discussed the state-of-the-art methods properly.
  2. The experiment/results should be compared with the state-of-the-art methods.
  3. Etc.

As a novice researcher, my questions are:

  1. What exactly does "state-of-the-art" mean in applied research?
  2. How to properly discuss or analyse the state-of-the-art method in a paper? Should it be discussed in the literature review section only or somewhere else too?
  3. In terms of applied research (like the application of data mining algorithms), how to compare the obtained results with the results of state-of-the-art methods? How to consider the results of state-of-the-art methods as a baseline and compare your own results with it?

Could anyone please explain with examples?

  • Could anyone please explain with examples? — Presumably you've read several applied research papers yourself. (Otherwise, how do you know what other people have done?) Use those papers as your examples.
    – JeffE
    Commented Sep 22, 2018 at 19:20

1 Answer 1


State of the art means the best known methodology. This should have turned up in your literature review and if it didn't you have a gap. The gap would likely be in finding results in very recent papers.

The first reviewer comment could mean that you haven't actually found and discussed the current best available method to compare with, or that you are somehow misrepresenting it in your paper.

But it isn't enough to discuss in in the lit review section. You need to explicitly show how your results compare with the results of the best available. Very possibly it would be a long section of a thesis, though page limits in papers provide a constraint. If your results are superior, you need to show how and discuss why.

As to number 3 on your list, the methodology you use is up to you to design and should be consistent with practice in your field, unless it introduces, also, a new comparison metric, in which case that needs to be discussed in detail as well.

For data mining, probably something along the lines of applying the "best known" algorithms along with your own to the same set of data is likely appropriate.

But you need to assure the reviewers that you compare apples to apples so that the results are directly compared.

Sorry, I don't have any examples.

  • 1
    It's worth adding that in most of computer science you probably have code, an implementation of this or that method. A proper comparison with prior work would apply different implementations (including your own, of course) to some problems and evaluating the results. A further question is, actually, what kind of a metric you should use to evaluate the results. Commented Sep 22, 2018 at 14:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .