9
$\begingroup$

I know this is not the fun topic most of us are here, but we need to get it out of the way, if we want to make it successfully through private beta. Consider giving the following a read: The Real Essential Questions of Every Beta

First and foremost, since my last question on the topic (Meaningful tagging with tag wikis full of meaning), we've been making progress. So thank you to all who have participated in that effort.

Now back to the dreadful topic of tags. There are currently more than 100 tags for less than 100 questions. That's not a big problem yet, because eventually there will be a lot more questions than tags. However, we need to be more zealous when creating tags and committing to useful guidelines on how to apply them. It is way harder cleaning these things up when they have multiple questions attached to it.

As a general guideline (and from a little bit experience) tags with fewer than 6 uses don't have a well enough defined scope. Some tags which are attached to too few questions will (iirc - don't lynch me if I'm wrong) also self-destroy after a while.

Here are a few examples I see (very) problematic, because their topic is not really clear or they duplicate in meaning:

  • : How is it significantly different from
  • : Most of us know what it is, but what questions should fall into that category.
  • : Do you mean molecular structure, or do you mean the mathematical category?
  • : Population analysis? Does it need it's own tag, or would pop actually be better?
  • : We have , there is (also a bit problematic), need a special one?
  • and me in harmony...
  • : I see the value in this, but it needs a clear scope and it shouldn't be redundant to
  • : Actually I'm surprise this exists, usually the name of the site is blacklisted as a tag; I guess the s was missing.
  • = ?

Meta-tags. They can be problematic and they are especially hard to handle in the long run. Such tags do not categorise the topic of the question, but the type of question. It may happen, that they group questions that have otherwise little to nothing in common in the same category.

  • : this is a popular one on science sites, but essentially you'd like to have every answer somewhat backed up with references. I use this as an example to clarify the concept. The tag usage is given and I think the tag itself is fine.
  • I know in some communities this tag is popular, be aware it can open the door for any kind of question. If you don't know what to type into the tag field, then you might start typing ques and then it'll pop up and populate like mad.
  • : Most questions here will be about software, so what's the point. Similarly
  • : You can scan almost everything, I don't think this is specific enough.
  • : Does that characterise a question?
  • : How does that relate to MM?
  • , Very meta, very redundant, almost covered by

I should point out that there is a chat room dedicated to discussions of tags: https://chat.stackexchange.com/rooms/107328/tags

$\endgroup$
  • $\begingroup$ Oops I went overboard and removed most of these tags. Guess I shoulda waited to hear back from others. In any case, I realize not that this is not so great to do in hindsight. It's better if we edit as we go so a bunch of old posts don't get randomly bumped. $\endgroup$ – Cody Aldaz May 3 '20 at 1:30
  • $\begingroup$ @Cody this is generally true, but we don't have old posts, yet. Thanks for the effort! $\endgroup$ – Martin - マーチン May 3 '20 at 8:18
  • $\begingroup$ Maybe software should be software-request $\endgroup$ – Cody Aldaz May 3 '20 at 16:54
  • 1
    $\begingroup$ @Cody That's pretty meta, too. I'd just include it in ref-req and hope that doesn't spiral out of control. The fewer tags you have to watch for misuse, the better it is for the long run. (Find me in chat if you want to talk more about it.) $\endgroup$ – Martin - マーチン May 3 '20 at 18:12
  • $\begingroup$ It looks like y'all are handling this - if there's anything I can help with, whether merging/synonymizing or giving advice, let me know. Like Drones, I'd recommend you consider spelling out stuff in most cases, avoiding abbreviations - particularly when they might be confused with something else, like (apparently) "scan". $\endgroup$ – Catija May 8 '20 at 14:10
6
$\begingroup$

It might have been noticed by some people, that Cody went through and implemented most (if not all) of your suggestions the other day. For example he removed the "functionals" tag from every question that had it.

It doesn't seem that anyone disagreed in any noticeable way, with anything that he did in this regard.

In preparation for our Private Beta, I also emerged myself in the Drones Private Beta for the 2 weeks leading to ours, and when they had tag issues, they came to agreement this way and in the chat, which Martin has pointed out, we have been doing here as well. I think we should continue to use our "tags" chat room, and also agree on changes via meta questions as they did on Drones SE.

I would also like to make some effort at answering the questions raised in Martin's question, about these:


: I actually do think a functionals tag is appropriate. DFT has been tagged 50 times, but many of the questions specifically have to do with functionals development, or choice of functional:

1) Appropriate functionals for prediction of NMR spectra of transition metal compounds
2) What are the best functionals for transition metal compounds?
3) What is the closest thing we have, to a "universal" density functional?
4) What makes PBE the most preferred functional over other GGA functionals?
5) Is there any DFT code or software that have option to use exact exchange-correlation energy functional?

While many of them do not really have anything to do with the functionals themselves:

1) What is the difference between BDFT and CDFT?
2) What correlation effects are included in DFT?
3) How to evaluate spin exchange parameters of a magnetic material using DFT?
4) Database or repository with values for the Hubbard potential U?
5) How established is density functional theory as a tool in drug design?

I propose the guidelines for using the functionals tag, would be that the question is about the development of functionals or choice of functionals. Everyone seemed to be using it this way anyway, so it did not seem to be a problem that "lost souls" would accidentally use the functionals tag inappropriately.


: The question was "what questions should fall into that category." So far all 5 questions with meta-GGA in the title or question body, would make perfect sense to me. I suspect we will be getting a lot of meta-GGA questions after I advertise on some mailing lists that I've been avoiding doing until we go public. The Minnesota functionals are some of the most popular functionals in existence and they are meta-GGA. The guideline for usage is simply that the question pertains to the use of the meta-GGA approximation.


: The question was "Do you mean molecular structure, or do you mean the mathematical category?" So far all 14 posts using the word geoemtry meant "material structure" (or molecular structure in some cases). We can make this clear in the tag's excerpt. I don't imagine we'll get a lot of questions about the mathematics topic of geometry, except those which have to do with the geometry of materials/molecules. I think most people coming to a specialized SE like this one, will not be using this tag in a problematic way (at least so far, 14 posts have used the term in the same way).


: The question was "Population analysis? Does it need it's own tag, or would pop actually be better?" and since only one question so far has had "Mulliken" in the question or answer, perhaps we don't need this tag (yet). Thankfully Cody already removed this tag from everything that had it. We can cross this bridge once there's enough questions with "Mulliken" appearing in it, and I agree that mulliken-population-analysis would be better than just "mulliken".


: Martin points out that we already have , and so far it appears that there's only two questions with CCDC in them, so the single question currently tagged with the ccdc tag can have it removed perhaps. Since it seems to have only been used once (in almost 150 questions), it may not be a problem?


: Thankfully it's gone now, and we have . But I think this should be dft+u not dft-u, and if "+" is not allowed, we can do dft-plus-u.


: Martin said: "I see the value in this, but it needs a clear scope and it shouldn't be redundant to ". I agree, quite a lot of DFT is KS-DFT, so if we have 1000 questions with the DFT tag and 900 of them have the KS-DFT tag, it seems a bit redundant doesn't it? The question is probably more for the followers of the KS-DFT tag (not me). Do they want to be notified all 900 times someone uses KS-DFT? Or only when someone is talking specifically about KS orbitals or the KS aspect of the problem is significant? This perhaps could be up to the followers of this tag to decide (I rarely use DFT so my opinion might be less important).


: Thankfully Cody removed it from all questions that had it.


: Cody already switched it to . I propose to keep it that way for now.


: Martin said: "If you don't know what to type into the tag field, then you might start typing ques and then it'll pop up and populate like mad." I don't know, does anyone here really type "ques" when they don't know what to type? We have only 3 questions out of almost 150, that have this tag, and I was impressed that people knew exactly how to use it, for example Mr. Pie from the MathematicsSE used it properly. This tag is on a lot of other SEs and it's going to keep coming back if we burninate it, so I'd say to keep it.


: Martin said: "Most questions here will be about software, so what's the point." Certain questions are specifically about the software, and out of almost 150 quesitons this has only been used (in my opinion very correctly) in 17 cases. The DFT tag has been used in 50 cases. Many questions are about the theory or approach to the modeling, not having anything to do with the software.


: Thankfully these tags were switched to software. Perhaps modeling-software can be a synonym so that anyone that comes here and types that ends up not making a new tag (like it happened the first time?). I think this was accidentally created because at the very beginning, someone thought that the "software" tag would be for far more than just modeling software.


: This was not a meta tag, it was for "strongly-constrained and appropriately-normed" functionals. Anyway, Cody removed it, which is okay because it only showed up once anyway.


: Again this is not a meta tag, it is for the physics concept of universality. The description correctly states "Questions related to universality classes of phase transitions."


: I think this was created at a time when there wasn't enough tags available to use. It's been used twice, both times in a reasonable way, and we can consider what to do with it if it becomes a problem.


, : Cody already switched everything with these, to reference-request. We can make synonyms if they keep coming up.


Summary: , , and are in my opinion good tags, and their usage can be described as I did above. If someone has a problem with one of these tags, perhaps we can have a Meta discussion about that tag specifically, as they did on Drones SE for 8 different tags. and have not been used much, so whether or not they are needed is something that we can decide once there's more questions (we have existed for only 8 days). should be in my opinion, although currently it has been switched to . The usage of can be decided by someone else, since my opinion on it is maybe not very significant. was removed from all questions that had it, and has been removed from all questions that had it. All the meta tags were in my opinion fine for now, except and have been switched to whenever they appeared, and synonyms can be made if necessary.

$\endgroup$
  • $\begingroup$ Our current user base certainly will use the tags quite appropriately. The steady base never really is a problem. It's almost always the drive-by questions. And if you are really aiming for 100/day, then you will have a lot of them. And every tag that is ambiguous and doesn't help your steady users, will lead to problems down the line. That is what I mean with soft-question. On Chemistry, wie have everyday-chemistry, which is misapplied about 95% of the time. We have the homework tag, which now says 'do not use' in all caps, but there are still people applying it. $\endgroup$ – Martin - マーチン May 7 '20 at 9:49
  • 2
    $\begingroup$ I am simply trying to point out a few things, before everything gets serious. I brought it up so that the community will deal with it. There has been really good progress thanks to Cody and others. Every community will run into Problems with tags, that is almost inevitable and there will be a certain flux to it. Keeping an eye on them is always a good idea. Few notes: I think functional-development is a way better tag than functional could ever be, and that could stand on its own. It's certainly a bit meta for the choice of functionals. $\endgroup$ – Martin - マーチン May 7 '20 at 9:57
  • $\begingroup$ Since you want to open this community to molecular modelers and materials modelers, I think the geometry tag is especially useless. Be verbose enough to let the tag say what it means. Keep it as a synonym, because in plenty of cases it will be applied anyway (esp. by the molecular community). || * All the meta tags were in my opinion fine for now [...].* Yes, it may seem this way, because right now you add them with a purpose. New users won't, and you'll always have to clean them up, otherwise they become completely worthless. It's a community effort, not a fun one. $\endgroup$ – Martin - マーチン May 7 '20 at 10:03
  • $\begingroup$ @Martin-マーチン I very much value all your insights and wisdom, especially since you have years of experience as a moderator on other SEs! I do think 100q/day is realistic eventually, but certainly not overnight. We have been at 13q/day for the first 2 weeks, and when we go public we will have far more users, but also fewer questions/person so I suspect we'll remain under 20q/day for a while (maybe the first year?). It is very good to be thinking ahead though, and I agree that spending our time wisely on tags now, can help save tonnes of time in the future! $\endgroup$ – Nike Dattani May 7 '20 at 16:55
  • $\begingroup$ I do agree that we'll get drive-by questions eventually, but I think the situation will be quite different to people's experiences on other science SEs: Physics, Chemistry, Mathematics, Biology, etc are all general subjects that people start studying in high school or elementary school. Materials Modeling is done by grad students and older, or 4th year undergrad thesis students. I'm not saying we won't get "homework" questions or drive-by questions, but I do think questions from elementary school, high school, and undergrads will be far less than on Physics SE: More will come from researchers. $\endgroup$ – Nike Dattani May 7 '20 at 17:04
  • $\begingroup$ I understand there might be a lot of drive-by "help me with this software" questions (not so much from high school students but more from researchers), & I'm trying to address that in the other Meta question. For "functionals" tag, I still do like it, but I'm starting to see your point about making it more descriptive. Let's see how things progress for now, since the tag has not been used at all since about 2 days ago when Cody removed it from questions that used it. In the beginning when there was almost no tags, people used it, but now they can easily get to their 5 maximum tags without it. $\endgroup$ – Nike Dattani May 7 '20 at 17:11
  • $\begingroup$ Sure thing, if uses of a tag are consistent, but another name eventually appears to be better, tags can be renamed (that is basically merging to an empty tag without keeping the synonym). When there will be mods, it's easy. So: caution - yes, worries - no. We'll see with time. And I'm happy to help while I can (wouldn't call it wisdom though). $\endgroup$ – Martin - マーチン May 7 '20 at 18:00
  • $\begingroup$ Since I have fallen into this situation by removing tags. I suppose it's useful to explain that I'm in the camp of less tags the better. Tags are only useful for highlighting questions to users. The minor difference of functional, meta-gga and density functional theory is pedantic and IMO not worth the effort considering what tags are useful for. I think we should establish a principle of less tags the better so that we don't need to have long drawn out discussions about tags as frequently and it encourages editing. I don't think every single tag needs to be rigorously debated (like this). $\endgroup$ – Cody Aldaz May 7 '20 at 18:30
  • $\begingroup$ I agree to go for fewer tags rather than more. I completely disagree that functionals and meta-gga should be absorbed within DFT, since DFT has 50+ questions and people that follow specific tags by email (like me!) won't want to be getting 50 emails/week. I would follow "functioanls" and "meta-gga" and instead get 10 emails/week. I agree that not every single tag needs to be rigorously debated. I think things are not bad right now, & what's most important right now is to encourage more user participation. I don't want our site to be closed due to the most common reason: "not enough activity". $\endgroup$ – Nike Dattani May 7 '20 at 18:36
  • $\begingroup$ I don't think one could properly distinguish functional from density functional theory. Or even if one could, that's no guarantee to me that it would be applied correctly. If we allow meta-gga there is like 4 or 5 other rungs of DFT which I guess would be ok. But what about all the rungs of other theories. The minor differences is again not that relevant (e.g. CIS, CISD, CISDT... CCSD, CCSD(T), CCSDTQ....) $\endgroup$ – Cody Aldaz May 7 '20 at 18:41

You must log in to answer this question.

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