1
$\begingroup$

There are many questions on our site that of the form (for example, this one)

Which algorithm/model can potentially solve this task X successfully?

This question is a bit problematic, because, unless you already solved the same or a similar problem, the best answer will probably be only an "educated guess" of the algorithm/model that would be appropriate to solve task X. For example, if someone is dealing with images, then convolutional neural networks may be a good first choice, but, who knows, maybe CNNs are not the best solution for this specific task, so this educated "suggestion/recommendation/guess" (which is not necessarily an "opinion") may not actually be appropriate, although, initially, it looks reasonable. So, this type of question is problematic because of what I just said, but it's a recurrent one, and, often, people need some guidance in order to solve a problem.

Assuming that this type of question is on-topic here, should we have the tags

for this type of question?

This would allow users to have the list of all such questions in one place, so that they can eventually search for a question that they also have before asking a duplicate one.

Or maybe we should use the existing tag for these questions too? I think that references are papers, books or any other source of information that can support something (e.g. a claim, quote, etc.), as the dictionary puts it

reference: the use of a source of information in order to ascertain something:

So, I don't think that algorithms and models are "references", so we shouldn't use that tag for these questions.

This question is related to this one.

$\endgroup$
4
  • 1
    $\begingroup$ Yep I think such tags make sense $\endgroup$
    – Dennis Soemers Mod
    Jan 27 at 18:33
  • $\begingroup$ @DennisSoemers In this case, which tag would you use? I think that in this case all of these 3 tags would be potentially applicable. $\endgroup$
    – nbro Mod
    Jan 28 at 11:58
  • $\begingroup$ Hrmmm I'd guess model-request, if that also covers things like architectures and not just requests for already-trained models with their trained parameters. Otherwise... algorithm-request? $\endgroup$
    – Dennis Soemers Mod
    Jan 28 at 18:03
  • $\begingroup$ @DennisSoemers Yes, I guess model-request, but sometimes it's not very clear, that's why I was wondering about your take on this one. $\endgroup$
    – nbro Mod
    Jan 28 at 18:44

You must log in to answer this question.

Browse other questions tagged .