2
$\begingroup$

We have a tag for and (currently) 16 questions. However, many of these questions don't have a primary arxiv tag of the form XX.XXXX. Our site policy currently recommends the use of an arxiv top level tag with each question, and an attempt to generalize this seems to have gone dormant.

In the specific case of clustering, would it be appropriate to use as a top-level tag ?

| |
$\endgroup$
3
$\begingroup$

Clustering is not always related to learning. If a question is related to learning, there is nothing wrong with tagging it with . Otherwise, I cannot see the point of adding tag to questions about clustering just to enforce some arbitrary rule.

| |
$\endgroup$
  • $\begingroup$ Only because we have this arbitrary rule in the first place, and there's been no movement towards revoking it :) $\endgroup$ – Suresh Venkat Sep 13 '11 at 22:37
  • 1
    $\begingroup$ @Suresh: I think that this is already a good reason to discard the rule of attaching an arXiv tag, which has never been enforced precisely because there is not always an appropriate arXiv tag. $\endgroup$ – Tsuyoshi Ito Sep 13 '11 at 22:42
3
$\begingroup$

I had checked the papers on arXiv about clustering. They use various top level tags depending on the topic (CV, LG, DB, ...).

It seems to me that most of the questions tagged clustering fall under LG, but the problem is that we have two tags and . It seems that on arXiv the second one is subset of the first, but we are using the first one in a more restricted way.

ps: having a tag specifying the subject area is useful. It organizes the questions so people can filter them easily, it is useful in the same way that AMS MSC, ACM CCS, and arXiv top level tags for papers are, and it is not arbitrary.

| |
$\endgroup$
  • $\begingroup$ “having a tag specifying the subject area is useful.” Yes, but it would become useless if we start tagging questions without considering the relevance of the tags just to enforce the “at least one arXiv tag” rule (like what Suresh suggested). $\endgroup$ – Tsuyoshi Ito Sep 14 '11 at 12:13
  • $\begingroup$ @Tsuyoshi, generally typical questions should naturally fall under some subject area. It might be the case that from time to time we have some question that is on a topic which is completely new and doesn't fall under any previous subject area. But that is an exception. Typically questions are related to something already published in articles/books and those are already categorized under some subject area, we can use that to categorize the question. $\endgroup$ – Kaveh Sep 14 '11 at 12:22
  • $\begingroup$ I understand, but what is the point of saying it to me? I am just claiming that Suresh’s suggestion to tag the questions about clustering mindlessly with [lg.learning] is pointless. Adding a relevant tag to each question is of course good. $\endgroup$ – Tsuyoshi Ito Sep 14 '11 at 12:24
  • $\begingroup$ Take as an example Suresh's recent question. at first site this doesn't look like a combinatorics question, but questions like this are studied in what is called analytical methods in combinatorics (and CS). It is not a perfect tag for the question, but I think it is one of the top level tags that someone interested in this type of questions will search for. $\endgroup$ – Kaveh Sep 14 '11 at 12:26
  • $\begingroup$ I still do not see your point. If the question is about learning, there is absolutely nothing wrong with tagging it with [lg.learning]. It is completely pointless to tag a question with [lg.learning] just because it is about clustering. I think that this is obvious. $\endgroup$ – Tsuyoshi Ito Sep 14 '11 at 12:29
  • $\begingroup$ ps: my point is that it is not stupid to try to find a relevant tag even when it is not a perfect one. If a typical researcher interested in the topic would search for that tag to find similar questions it is fine to add the tag even if when treated as a concept it might not fall under that tag. The point of tags is pragmatic and helping people filter, not ontological, i.e. the question we should ask when tagging a question is not "is this question really a learning theory question?" but "will a typical researcher looking for this type of question filter questions using this tag?" $\endgroup$ – Kaveh Sep 14 '11 at 12:31
  • $\begingroup$ Well, do whatever you like. I do not care much about tags. Even if tags become completely meaningless, that will not mean that the website becomes meaningless. $\endgroup$ – Tsuyoshi Ito Sep 14 '11 at 12:31
  • $\begingroup$ pss: sorry for unclarity. You replied in the middle my reply (the first and second were supposed to be together). I think we should use our area tags in a similar way that subject classifications (ACM CCS, AMS MSC, arXiv top level tags) are used for papers. It hasn't made those subject classification systems meaningless. $\endgroup$ – Kaveh Sep 14 '11 at 12:33

You must log in to answer this question.

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