MO editors' lounge

8:28 AM

Hi @RicardoAndrade, I read your last comment in the post : "I have a silly question. Is there any way of knowing if this feature request is actually moving forward or not?". In my opinion, what Scott suggests is a good idea, but is it possible ? Do you have some news ?

Ricardo Andrade

9:21 AM

@Sébastien: I had the same question myself, as the feature request system is very opaque. As far as I can tell, us mere mortals have essentially no way of knowing whether a feature request is moving forward or not, barring an answer from a mathoverflow developer.
So I have no news, really. I can say, however, that it seems extremely unlikely that this feature request will go through. Similar features have been requested on meta.stackoverflow before, and were duly rejected.

@Sébastien: Apparently, all we can do is vote it up, and hope that a response comes eventually.

quid

12:51 PM

@SébastienPalcoux yes this is a bit tricky situation. One way to deal with it is to acknowledge right away in the comments the need for modification and to provide any clarifcation needed, but no to update the post itself each time something comes up or one notices something by oneself; only sometimes then doing several things at the same time.

quid

4:49 PM

@RicardoAndrade my usage of 'keywords' was not very good as in the usual sense they are also rather meta-information. What I meant is that one can think of tags in two ways: on the one hand one sees them when the question is in a general list of question, say the frontpage. on the other hand, one can use tags to create specific lists and to filter by tag and so on.
By keyword-usage I meant the former. That is the tag is somehow a flashy sticker put on the question that conveys something on its content. By the meta-data I meant the second use case. Of course there also the tag conveys some m…

I meant it is no problem, not now a problem. Somehow I cannot edit this as it got too long, I think. Not sure why I even managed to post it.

For the tag-tree. I basically meant just a structured list of the existing tags. Or maybe a graphic conveying this. But really a structured list I think would also do, just like the MSC list. Say, top-level nt.number-theory then all then listing all the general/large numbertheory tags, algebraic-nt, analytic-nt then below analytic-nt, prime-number-theorem, l-series, and so on.

Just so that some number theorist can even get somewhat convenietly an idea of the relevant existing tags without reasing through the list of 1500 or something and growing number of all tags.

Also, at the moment the top-level tags are hardly documented anywhere. And, yes, for some it is a bit unclear what the top-tag is. In theory for linear-algebra there is a default top-level tag, namely ra.rings-and-algebras (in the description of the category on arXiv lin-alg is mentioned.)

But, the usage of this tag feels (to me) somewhat strange for certain questions tagged lin-alg. So, that I do not add it to all the lin-alg question without top-level tag I happen to note, either.

My and I think I was not alone feeling was for a long time that it is still better not to start to tinker with the top-level tag list since if one starts there will be never-ending discussions what should or should not be top-level.

This is a bit unstructured but I hope it clarifies a bit what I meant.

Philip van Reeuwijk

5:19 PM

quid: you are right, I agree with you. I will refrain from participating in this discussion too much, since I don't have enough points to edit posts yet so tagging policy basically is none of my business. But let me just remark that it would, in my opinion, be great (if probably extremely hard to accomplish) if this "tag"tree

that went wrong

I meant to say: if this "tag tree" would work in such a way that high level tags would actually subsume low level tags, so that e.g. filtering by "nt.number-theory" would also show all questions tagged "analytic-nt".

In this scenario using more than one tag would only be necessary to mark truly multidisciplinary questions, instead of conveying logically superfluous information just to make it easier to find the question (which, given the current tagging scheme, is quite legitimate but shouldn't be necessary). Anyway, I'll leave you actual editors to it, best of luck!

Yemon Choi

7:31 PM

I am not sure I will participate much in this chatroom, but I just wanted to ask@quid the rationale for retagging this old and dead question mathoverflow.net/questions/79960/…

Ricardo Andrade

8:26 PM

@quid: Thank you so much for the clarifications. That really helps.

@Philip van Reeuwijk: Welcome. Even if you cannot edit, good ideas are always welcome! That is certainly an interesting idea for simplifying the tagging of questions! I wonder if it could be made to work with the tagging on stackexchange.

@Yemon Choi: Quid does not appear to be around right now, so I will try to answer for him. I think that question was edited to get rid of the tag 'algebra': see meta.mathoverflow.net/a/689/21095 for the request for deletion of the tag 'algebra', which no longer exists now.

By the way, welcome Yemon! Your participation here will be greatly appreciated, if you decide to enter the conversation.

François G. Dorais

8:42 PM

Re: tag structure. That's the wrong way around. Each user has a different view of the tag ecosystem and they will keep doing whatever they think is best according to their view. Imposing more structure is almost hopeless. A much more promising approach is for interested and capable users to develop applets that will sift through tags according to some kind of clustering algorithm to give the users the questions they are interested in with much greater accuracy than the raw tags.

I've talked to a bunch of people who work in data mining. They suggested it is now realistic to have a real time system to suggest relevant tags before users post their questions. Sadly, nobody ever volunteered to explore this possibility seriously.

Ricardo Andrade

@François: Regarding your statement "Each user has a different view of the tag ecosystem and they will keep doing whatever they think is best according to their view." I could hardly agree more. That is true for tags as it is for so many other things in life. Partly (but not entirely) for that reason, the tag system appears to be confusing for everyone. Well, perhaps I am over-generalizing, but it certainly confuses me a great deal.

Also, @François, would it be possible to integrate such hypothetical tag related tools into the stackexchange software?

quid

9:31 PM

@YemonChoi basically Ricardo Andrade answered it already. Over the last weeks I tried to get rid, in a reasonable way, of the tag algebra that was officially deprecated a while ago. To this end I retagged all question tagged in this way where the tag did not feel redundant or unecessary. The one you asked about was the last I retagged (or at least among the last three of about 60 total). In fact I hesitated to retag this one precsiely for the reason for which you are presumably asking.

@YemonChoi Yet then at the very end I decided to still retag this one too. Basically for consistency of the approach and convenience. In this way I could say in my request for the deletion of the algebra tag that all still having it are not that badly tagged without it (all having a toplevel tag or linear-algebra). Instead of saying this, and adding except for this one q, but it might not be necessary for this question to be properly tagged.

@YemonChoi Or, in brief, it was done as part of a larger task. I agree this one retag in isolation would have been bad. In the above context it made somehow sense to me (though I still considered avoiding it). I hope this clarifies the situation. I also hope to see you more frequently here! Perhaps not only if you disagree with me, but even this would be fine with me. :-)

@PhilipvanReeuwijk This is an interesting idea. One could perhaps implement this somehow in a similar way to tag synonyms but without really declaring them the same in the process. So one would do this manulally and could only do it in clear cases.

François G. Dorais

10:08 PM

@RicardoAndrade: Yes, we would try to integrate a tag suggestion tool into the software, with MO as a test site.

Transcript for