27 December 2011

Tagging links: a hint


One of the more popular classes of productivity app is the social bookmarking app.  The principle behind sites like deliciousdiigoGoogle bookmarks, etc. is that a group of people will tend to reuse the same (or similar) bookmarks for similar items thus leadings, eventually, to a self-organizing collection of links.

Whether the hypothesis is a good one is, for my purposes here, irrelevant. (Personally, I think it is false because too many people seem unable to identify key and rather obvious tags associated with relatively obvious web resources.  But that's another story.)  These sites are also very useful for individuals simply trying to keep track of web resources for themselves.  I used to have a home-made system that treated keywords and tags as the same, which seemed to lower the number of tags one actually needed.  (I've not kept that system up just because I haven't time to nurse the code.)  Still, I find that existent systems are good enough - you probably will too.

But the question remains, how does one use tags in a sensible way?  This is a broad question that I don't really want to get into because it requires one to eventually call forth the cumulative knowledge of the library sciences with respect to categorizing resources.  And I know just enough to know that I don't know enough about that to comment intelligently.

There is one small "trick," though, that I've developed over the years, and that seems to work quite well for me.  It is this trick I'd like to share.

The trick involves trading off two goals: minimizing the number of tags one must remember and apply to a given resource; and capturing sufficient meta-data to be able to efficiently search for resources later.  I'll show the trick with and example.

Let's say you want to tag a resource as relating to decision making.  Depending on which site you use the syntax will vary, but most of the good sites allow multi-tag words or some approximation thereof.  I'll use double-quote to denote a multi-word tag, and I'll italicize tags to set them off from surrounding text.

There's three general techniques by which you can tag that resource.
  1. decisionmaking - jam multi-word tags into a single word, which has the merit of reducing the extra characters (underscores, spaces, quotes, etc.), but hurts readability.  And what about makingdecisions?  Does that count?
  2. decision-making or decision_making or "decision making" - You can separate words with special characters.  Some systems let you use hyphens; others support the use of double quotes to group multi-word tags.  I like hyphens because I don't have to hit the shift key to get one.  Other systems only allow underscores to join multi-word tags.  I find that annoying, but the underscore is a time-honoured mechanism for forming multi-word variable names in many computer languages - and since many of these systems started as some programmer's pet project…. Well, you know.
    Finally, some systems, like diigo, separate all tags with commas and allow spaces in tags.  This is perhaps the best, most natural, and efficient way of specifying multi-word tags.  That's one of the reasons why I have for a long time now used diigo.
  3. decision and making - The last technique is perhaps a little counter-intuitive, but this is in fact my trick: treat each word in a multi-word tag as a separate tag.
Why on earth would you treat a multi-word tag as a whole bunch of tags?  Doesn't that break the semantics of the single, multi-word tag?

Well, yes it does, on the surface.  But I think that's a result of limiting your thinking to just what tags are most appropriate in some objective way.  Rather, I think we need to think of how you (and possibly others) might search for resources based on those tags.

So, what simpler way is there to search for something than just typing all the tags as separate words?  And what is most likely to catch related resources that might also interest you?  I think it's the single-word tag approach.  It's also pretty efficient in the long run.

For example, if you search for decision in a collection tagged as I suggest, you'll find all the resources tagged with that term, including those tagged with decision and making.  On the other hand, if you had treated decision making as a single tag, your search for decision wouldn't find them.  If you don't know that decision making is a single tag, then you might not find anything useful.

Many systems will show you a list of tags that occur with the tag for which you're searching.  So a search for decision will show, among other related tags, making.  Those systems will also show you other related tags, some of which may trigger new ideas.

This means that what you're really doing is constructing an implicit taxonomy of terms, where, in this case, decision is the most general term, and decision making is an item under the taxonomic tree for decision.  And it all happens naturally without any effort on your part.

So, as far as I can tell, treating multi-word tags as lists of single word tags is one of the easiest ways to enter tags, one of the most natural ways to search tags, and self-constructs a taxonomy of meta-data along the way.

Win-win-win, right?

No comments:

Post a Comment