What kind of user-level abstraction should I use for hierarchical tags?

Question

By user-level abstraction I mean what should I call them and what kind of icon should I use to represent it in the UI? The concept of tags should be familiar to most users right now, but I'm not so sure hierarchical tags are, nor does the abstraction of tags completely fit.

What I have is actually something between folders and tags, in that each subtag is a proper subset of its parents. To get the contents of a folder-tag-thing you take its contents plus the contents of its sub-folder-tag-things and do that recursively. In a way they're like (and here's where I just partially answered my own question in the process of asking it) categories. What do you guys think of calling them categories? What kind of icon would you give it, since a category is an abstract concept rather than a physical object that can take a shape?

(Can wiki if requested)

Edit: For clarification, what I'm looking for is an abstraction for an end-user like Aunt Sally that easily grasp the concept. Ideally there is also a graphical representation (an icon) that can be easily associated with that concept.

Edit2: One thing I did forget to mention is that an item can exist in more than on category (much like how Google Doc allows you to add an document to multiple folders). I guess I've kind sold myself on calling them categories. It just fits everything you can do with them. Something can belong to a more than one category, subcategories make sense, most subcategories people normally create lend themselves to an is-a relationship (for example, a Windows user might have say a folder called resumes in their My Documents folder, because a resume is a document, it would make sense that it's in My Documents as well if you think of the folders as categories)

What I still haven't figured out is the icon (I'm 99% sure I'm going to be using some kind of TreeView to display them in) I could just use a folder icon I suppose, maybe a custom folder-like icon unless anyone else has a better idea. Google calls them folders, so I guess it can't be too bad right?

To clarify further, my target audience will be people technical enough to know how to download and install the app, but that's about it.

Answer 1

It sounds like an outline to me.

There's the Outline Markup Language (OML).

Answer 2

You could use "bundles" (as del.ico.us does). I am not sure if more than one level is meaningful for tags though.

Answer 3

How about a tag "detail"? Since a subtag is a proper subset, it can be thought of as applying detail to a tag...

Answer 4

If your target audience is technical then TreeViews and/or Folders are common ways of abstracting hierarchical data. They're all variants of the Composite pattern.

In the post-google era though, I've tended to consider hierarchical categorization less important than a flat tag/label space combined with a very fast search. IMHO if the size of the dataset is large (e.g., > 1000 documents/elements) then the overhead of hierarchical categorization overwhelms the nominal benefit of having documents placed into nice neat categories.

If your target audience is non-technical then it's been my experience that non-technical users do not get hierarchies and won't take the time to to figure out more than a single level of categorization.

Answer 5

For me, the answer comes from the Opera Desktop Team blog (as well as Dotclear’s blog, which used the same convention and thanks to whom I discovered it too):

tag:subtag - CSS:font

When you click on the CSS:font, you know it is part of a larger CSS tag. The tag page should have a text indicating that a parent tag is available (or you can find it by yourself, much on Wikipedia with direct URL manipulation).

I have a discussion running on WordPress.