3

I recently posted a question here about how we should handle tag hierarchies on this site.

I pointed out the many ways in which are current use of sub-tags is inconsistent, with many of our sub-tags being added to only a tiny fraction of the questions that they would appear to apply to:

I'm certain there are many other examples.

So far, the responses to my question indicate support for continued use of these tag hierarchies.

The top-voted policy indicates:

Tags should "drill downward" from the largest possible catchment to more specific tags with Series tags used to cover an entire canon universe

and

Sub tags should be used to reflect specific topics that are likely to have a reasonable level of interest. Any tag that's only had one question "hit" in say, 6 months should be considered for closure or merger.

The problem with this is people don't use these tags appropriately.

Any utility of having more specific tags hinges on the tags being used. It seems pretty clear that we're not talking about dozens of instances of sub-tags that need to be added.

We're talking thousands.

At this point, the only way I can see us getting a handle on these is by having a script automatically append selected sub-tags based upon appearances of that tag in the text.

For example, add to any question that has the word "jedi" in it, or add to any question with the word "Voldemort" in it. We may wind up with some false positives, such as questions that mention Voldemort, but aren't about Voldemort (although he is called "He Who Shall Not Be Named" for a reason!), but the cleanup of the false positives should be significantly easier than our current situation.

Similar scripts should probably be run whenever a new sub-tag is added, since the consensus seems to be that they should only be added once a certain level of popularity is demonstrated, which means that there will be older questions lacking the tag once a decision is made to add it.

Even if we can't get these automated, we need to identify which questions are lacking sub-tags, and start figuring out who/how/when these will get fixed.

  • Related: A proposal for tag hierarchy on SO @Meta.SE. That would solve most of these problems from systematic standpoint – DVK-on-Ahch-To Aug 1 '14 at 17:03
  • @DVK I like your proposal, but I don't see how it would solve our current issue. The problem is that the more specific tags aren't used on most of the questions that they're applicable to, and your tag hierarchy would seem to address the opposite situation (questions tagged with the more specific tags missing the more general category tags). – Beofett Aug 1 '14 at 19:42
  • 1
    having a script automatically append selected sub-tags - And if we remove them, what prevents a later run from re-adding them? – Izkata Aug 1 '14 at 22:54
  • @Izkata good question. Perhaps the scripts would be one-time jobs. Personally, I think these types of sub-tags are fundamentally unmanageable, which is why my answer on the linked question proposes we do away with them, but so far support for that is practically non-existent, so I would like to start planning the details of how we'd make sub-tags work. – Beofett Aug 1 '14 at 23:23
  • 1
    I think some sub tags are worthwhile, and others aren't. If the justification for applying a tag is that a certain word is used in the question, the tag isn't necessary. However, there are examples like the wandlore tag for Harry Potter where the majority of questions with that tag don't include that word in their text anywhere, making it useful for classifying and searching for questions on that topic. – Anthony Grist Aug 2 '14 at 16:48
  • @AnthonyGrist only if the tag is applied consistently. – Beofett Aug 2 '14 at 16:49
  • 1
    @Beofett Which we're all more than aware isn't currently the case. I think we're all also very much aware that regardless of what we do now, some - even most - people aren't going to use those tags when they should. We either accept that and commit to doing the massive amount of work involved in cleaning up our tags anyway, or we end these discussions now and see if people care more about the topic in a couple of years. – Anthony Grist Aug 2 '14 at 16:59
  • @AnthonyGrist I'm not sure who you mean by "we're all aware". The handful of people who have read this post? The slightly larger handful that read the other post? Either way, I'm pretty sure it's not a majority of users on the site. Regardless, the whole idea of either committing to do the cleanup or not is the point of this meta discussion. – Beofett Aug 2 '14 at 19:40

You must log in to answer this question.

Browse other questions tagged .