« Semantic web apparently moving along | Main | After Web 2.0? Web 2.0 2.0 »

Technorati tags as metadata: making them more meta

Is it really metadata when you have to announce "here are my posting's Technorati tags"?

More bloggers are embedding Technorati Tags into their postings, and it's great to see a major league app use user-created metadata for anything. The popular convention of announcing what your Technorati tags are as part of your blog's content, though, made me wonder: can we add these tags where the casual reader doesn't see them, so that they really are metadata instead of being additional tagged content? For example, if I write a posting about RDF and XMP without actually using the word "metadata", how can I get a Technorati tag search on the word metadata to find that entry?

A View Source on my last posting here will reveal some experiments I did with Technorati tags that had no content between the a elements' start- and end-tags. (Experiments elsewhere showed that a single-tag empty a element got treated as a start-tag with no end, so that the text after it was underlined as if it were a link anchor.) Of the following three a elements, the first resulted in an entry at http://technorati.com/tag/bobtest4 and the third in an entry at http://www.technorati.com/blogs/bobtest6 (along with the comment "Hey Bob! Is your blog about bobtest6?" on the right):

<a href="http://technorati.com/tag/bobtest4" rel="tag"></a>
<a href="http://technorati.com/tag/bobtest5" rel="directory"></a>
<a href="http://technorati.com/tag/bobtest6" rel="tag directory"></a>

Considering that a rel attribute is supposed to hold "a space-separated list of link types," and that "White space characters are not permitted within link types," I would think that the second line above would put a bobtest5 entry at http://www.technorati.com/blogs/bobtest5 and and that the third would put entries at both http://www.technorati.com/blogs/bobtest6 and http://www.technorati.com/tag/bobtest6, but it looks like Technorati's crawler treats "tag directory" as a single rel entry and ignores the one with a value of "directory".

I could be wrong; this is just guesswork based on experiments. (I've searched for further background from Technorati on how this really works but had no luck. Although I signed up for their developer's program and sent various e-mails, I still have no idea know how to subscribe to their mailing list.)

A nice bonus is that the bobtest4 example sort of works from an RDF/A perspective, too, although a predicate of "tag" is awfully vague. Even a name like "ttag" would give a better clue as to what kind of tag we're talking about: a Technorati one.

In this and future postings, I'm grouping such metadata inside of a div element with a class value of "technoratiTags". Does anyone have ideas for a better way to incorporate empty a elements with Technorati metadata into XHTML, particularly if it can take advantage of other metadata standards work?

(More writing on the use of Technorati tags: Big-time app uses the a/@rel attribute, boosting "folksonomy" development and Folksonomy tags for indirect linking).

Comments

(Note: I usually close comments for an entry a few weeks after posting it to avoid comment spam.)

Doesn't Technorati also use atom:category or dc:subject elements for determining a story's tags? I think this is what you really want, rather than putting empty anchor tags on your pages:

If your blog software supports categories and RSS/Atom feeds (like Movable Type, WordPress, TypePad, Blogware, Radio), just use the included category system and make sure you are publishing RSS/Atom feeds and your categories will be read as tag.

Also, and this is the best part, in my opinion:

You do not have to link to Technorati. You can link to any web page that ends in a tag - even your own site!

So a regular link to Wikipedia, or your own categories, or whatever, can have rel="tag" on it, so you don't need to artificially link.

MovableType puts a block of RDF metadata in the entry, but comments it out. I do see dc:subject="metadata" in there for this entry, but nothing showing up at http://www.technorati.com/blogs/metadata?sort=recent, which is where I'd expect to find it, so I don't think that Technorati is parsing anything inside the comments. I do see something at http://technorati.com/tag/metadata, but I think that comes from the kinds of tags I describe in the posting.

I did see the part about any URL working as long as you had rel="tag" in their documentation, but if that worked then my zamfir10 test at http://www.snee.com/sneetard/2005/12/december_13_test_2.html should have worked. The use of atom:category should work, but MovableType only lets me pick one category and (I think) converts that to atom:category in the Atom feed, and I'd like to assign multiple bits of metadata to a single posting.

thanks,

Bob

Hi Bob, have a closer read of the http://microformats.org/wiki/rel-directory spec page.
Using rel="directory" asserts that the linked page is a directory listing containing an entry for the current page.
However, the Technorati blog directory is a tagged one, and so you should assert that it is a tag directory fro your page, if you want us to include it.
We do combine the meanings - rel="tag" says 'this is a tag'; rel="directory" says this is a directory for the page, ie the blog as a whole (indicated as the homepage) What we assume from a bare rel="tag" is that the context is the blog post. The scope of rel="tag" is kept unspecified deliberately, as it is used in multiple contexts (se xFolk and hReview, for example, as well as blog posts). Adding rel="directory" resolves this ambiguity.

Hope that helps.

PS Putting empty anchor tags is bad practice - if you care enough about the tagspace to use it you should link it, and your links may confuse your readers (especially those using screen-reading software) and search engines that are not aware of these microformats.

Thanks. I must admit that my example of tagging a post that doesn't use the word "metadata" with the keyword "metadata" is an edge case. Tagging existing content words as keywords shouldn't be that challenging, and I'll try to get into that habit.

Bob