Sorry if this counts as "navelgazing meta-discussion", but I'm actually wondering because part of a system I'm building for a tool of mine will be intimately driven by user tagging, so I'd like to learn how the tagging system works around here for reference. The exact workings of the system are immaterial, but you could think of it like tagging Twitter-like posts, or something to that effect (sans the juvenile post material).

Actually, while I'm at it, I have a few questions:

  1. Why exactly would keeping my tag set as conservative as possible be preferable to allowing expressive use of tags? I'm under the impression that the more tag data provided, the better. More specifically, would this be something implementation-specific (like the difference between a social bookmarking service and, say, Stack Overflow)? Or is there a specific reason to want to follow this strategy in general (like to make a taxonomy)?
  2. Would implementing a thesaurus-like system to minimize unique tags be worth the effort? Or is it just easier to pimp slap people into using a standardized set?
  3. How would a standard set get to be defined? I dun wanna do it all myself, but I don't want to argue with people how to get it done either if it ever became important to do so. Am I just going to have to pick which path I don't want to do the least? Or is there some 'correct' procedure I could implement or borrow for picking tags (and keep in mind I'm talking about tagging in general; not programming-specific tags or anything like that).
  4. What would be a good way to form a taxonomy from the existing tag set if I needed one?
  • @Daddy, from the phrasing of your question it sounds as if you're planning to provide tags to your users--which is exactly the opposite of the way that tags are done here. So I'm not sure I understand your question. Commented Oct 21, 2008 at 13:06
  • No, I mentioned that I don't really want to have to do that. That could mean a huge amount of work for even several committed individuals. Commented Oct 21, 2008 at 13:08
  • Unless, of course, I enforced an ultra conservative policy for the available tag list. Commented Oct 21, 2008 at 13:09
  • 1
    @Daddy Warbox, please consider (and I'm no one here on SO to request this, please note) reposting this question in a non-Stackoverflow Centric way... oh no, wait, then it would get closed as not-programming related. I'll post a link if I manage to remix your question... Commented Feb 24, 2009 at 11:19

2 Answers 2


For tagging systems in general, the type of tagging you want depends on the nature of the media being tagged.

If you have a media that is difficult to index, like images or video, then you want a loose tagging system that encourages putting as many tags as possible with each item. The more tags the merrier, because you will need to use the tags to help index the content for searching.

For something is more easily indexed (text!), you want a much more rigid system where it may take an extra step or two to create the tags. This is because in this situation, the items themselves are the tags in a manner of speaking. You don't need tags to index the content, so the user-added tags will be used only to categorize it. And to be useful for categorization you need to make sure that multiple users are putting things into the same categories.

  • Ah good point. Yes I'm using text, so that could be a good reason to keep it conservative. Commented Oct 21, 2008 at 13:12
  1. Lots of tags are great but you need to make sure that duplicates and typos are cleaned up. Here, the people with enough rep-points tend to clean up tags (I know I do) on other people's posts and I've even gone through the tags section just to clean things up. It's time consuming but necessary.

  2. You certainly could try and create an automatic thesaurus (or a "Did you mean...?" on tag entry) system but it's not going to be as accurate as human intervention.

  3. The standard set here is created by the users. Tags are reused based on new questions' authors picking them and people adding/editing them later on. I don't see why you'd need a particular starting point but make sure you edit your initial data set so they're using enough tags.

  4. You can't really -- not automatically, at any rate... Tags are folksonomic, designed to be plentiful and overly descriptive through organic input. Taxonomic descriptors are deliberately wide-reaching so many things can fit under a section.

  • P.S. I was afraid of that (#4). Though I have heard of people making systems like that... Commented Oct 21, 2008 at 13:34

You must log in to answer this question.

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