2

I see that every paper which propose (present) something (new) (over existing methods) should possess an experimental result section and some tables and figures and numbers. Some papers may be rejected if they don't follow this format regardless what they are going to say. They may are needed for the evaluation of the method or comparison of it with the related methods.

I myself, when read some papers, based on the arguments in them, propose a new or more general techniques which covers their shortcomings. Or I may propose a new formalism or tool which can be shown that is novel or useful.

However, providing some experimental results for them is not always easy. Some measures should be defined and samples should be gathered to get some statistics ( for something that you know works in practice). In fact some times, I don't know how can I offer such results or whether it is necessary or not.

Then, I would like to know which type of researches need those quantities and tables and diagrams?

Or in the other words, which articles don't need a results section? How can I write or put my arguments to show that it doesn't need further evaluation.

I give some examples to show my problems (first I should say my filed of study is computer science):

  • There are some grammar or annotation scheme to build a linguistic corpora for English, there is no such corpora for Persian, I justified a similar scheme for building such a corpora for Persian based on the language features, Should I provide (experimental) results?
  • I developed a software which filter a webpage and remove noise and reformat the content to enhance readability and comprehension for non native English speakers ..... I used the results of another article on reading comprehension which didn't implement a web tool for their foundings.. Should I provide (experimental) results?
  • There is an algorithm for converting a linguistic corpora specially designed for English language, I generalized it to cover some other languages and justified that with some arguments, should I provide results?
  • I proposed a formalism for extraction rules for a data extractor from webpages, I provided many examples how it works, and how it is more robust, understandable and easier to use and based on previous tools and methods and how it borrowed their features ..., should I provide (experimental) results? ......
8
  • 1
    Are you talking about journal articles or about funding proposals?
    – jakebeal
    Jun 5, 2015 at 12:36
  • @jakebeal Journal articles
    – Ahmad
    Jun 5, 2015 at 15:34
  • It's unclear to me what you think the difference between an "example" and an experimental result is in this context. But the basic answer to your question, is yes, you should provide results that prove your methods are better than what is currently available.
    – Bill Barth
    Jun 5, 2015 at 16:38
  • @BillBarth based on my question, then I get that the answer is "YES", whenever someone propose something over the existing ones, should do some experiments.
    – Ahmad
    Jun 5, 2015 at 16:45
  • 1
    @Alexandros i just mention it as an example, you can suppose my software has unique features which enhance comprehension for non natives. However I admit that need an experiment to show it, but I used the foundings of some other articles in language comprehension and developed a tool for it
    – Ahmad
    Jun 5, 2015 at 21:35

2 Answers 2

6

In all the work you are giving as examples, you are essentially developing methods. Not giving any experimental or very strong theoretical evidence that those methods do what they are supposed to do has many negative effects:

  • It strongly diminishes the value of your work to the extent that you may have trouble publishing work that you could otherwise publish it in a high-ranking journal. Peer reviewers may very well demand that you report on the results of applying your method to some data and may reject your paper if you have no very good arguments for not doing so.

    Publishing methods without giving such evidence is the exception and usually only done, if deriving and explaining the method makes for a whole paper and applying the method is done in a follow-up paper or if the method paper and the experimental paper are better published in separate journals. And even then, there is usally some short application to (often artificial) example data in the method paper.

  • It makes it much less likely that people ever use your work, build upon it or cite you. If I read a paper in which the author proposes a method but fails to show any application, I would strongly suspect that this is because they can’t, which in turn hints at something being very wrong with the method. I would be very skeptical about such a method and would be much less likely to use it.

  • It will be harder to get funding for your work or justify that you spent your funding well.

  • Whether you publish it or not, applying methods is one of the best ways to learn what demands and problems occur in reality, which allows you to develop better methods.

  • If you repeatedly do something like this, you may obtain a bad reputation due to the above.

How bad all this is, eventually depends on your field’s openness to theoretical work and how difficult its is to apply the methods and properly evaluate them. But as far as I can tell from my limited insight into the computer-science literature, I would consider the impact to be rather strong in your case.

Given the nature of your work, I expect it to be impossible to provide convincing theoretical evidence that your methods work. For example languages are notorious for exhibiting weird behaviour and having tons of exceptions. Thus only by applying your method can you demonstrate that your it isn’t rendered useless by peculiarities or special cases you did not account for.

So, to sum it up: Yes, it may be possible to publish your methods without experimental evidence that they work, but I would strongly advise against it.

If you strongly despise experimental work, I suggest that you team up with somebody who does this work for you.

1

I think that you answer your question yourself in its title: experimental results section is needed in papers, reporting on experimental research studies. However, not all studies with quantitative results are true experiments. Thus, speaking more general, all quantitative research studies are expected to report their results by including corresponding quantitative results sections.

10
  • I mean for any proposal, they ask for results! are there proposal which don't need them? Maybe I should study type of researches!
    – Ahmad
    Jun 5, 2015 at 12:02
  • 1
    @Ahmad: A proposal by definition cannot have the results section, since there is nothing to report (study has not been performed yet at the time of submitting proposal). However, a proposal IMHO can and should include the expected results section. Jun 5, 2015 at 12:08
  • 1
    A "proposal" is typically an ask for money from some organization, and a "paper" or "article" is a document reporting on some results of research. When a paper says "we propose..." the authors are usually being a bit demure. They don't want to be too boastful, so they say "we propose this new theory about how birds fly". It's not a style that I like, but it's very, very common. They then go on to provide some sort of evidence backing up this "proposed new theory". They may be experimental results, a mathematical proof, or other evidence.
    – Bill Barth
    Jun 5, 2015 at 12:18
  • 1
    @BillBarth: I guess, your comment is addressed to the OP. If I'm right, consider mentioning him, using @, so that he can be properly notified. Jun 5, 2015 at 13:41
  • 1
    @Ahmad, it depends on what you are trying to prove. If you have a theory about why Arthur Conan Doyle wrote Sherlock Holmes in a particular way, then that evidence might be quotes from the books, Doyle's letters, other published criticisms, etc. What evidence would convince you that what you want to say is true? That's a good place to start. You can't just go around claiming things are a certain way, you need to convince people. In math this is a traditional proof that goes from proposition to proposition, etc. In other fields it may be something else. What field are you studying?
    – Bill Barth
    Jun 5, 2015 at 15:59

You must log in to answer this question.

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