The W3C just announced “Extending XLink,” a Working Group Note that “describes changes that could be incorporated into an XLink Version 1.1 specification to address usability, dependence on annotations provided by external grammars, and interoperability.” It’s short and makes a good beginning at what I see as the fundamental problem with moving the W3C’s little-used linking standard forward: everyone agrees that XLink is too complicated and messy, and should be cut down to a leaner, more manageable core, but few agree on what to cut out and what to leave in. (It’s a classic engineering problem, and anyone who appreciates it should be even more impressed with the effort that went into whittling down SGML into XML.) The new note addresses the simplest, most basic problems to clear out of XLink. I’d summarize them here, but the document is so short that if you’re interested in linking at all you should take five minutes to read the whole thing yourself.

(One picky tech writer complaint: it assumes that all readers know what the acronym “IRI” refers to. Try a Google search on the term for a laugh—International Republican Institute? Information Resource, Inc.? International Research Institute for Climate Prediction? Those are just the first three hits, with the W3C/IETF meaning of the acronym currently showing up at number 48. The XLink note refers to Internationalized Resource Identifiers, a complement to URIs that reduces the dependence on ASCII.)

On Monday I wrote about Technorati’s use of the a element’s rel attribute to let people link weblog entries to taxonomy entries. I mentioned that this attribute, which was designed to implement link typing, has been all but ignored in its twelve-year history, and that its use by a big-time application would give people more incentive to use it.

Yesterday one of the biggest applications of all announced a use for the same attribute. To fight the practice of referrer spam (the addition of irrelevant comments to a weblog entry in order to boost the number of links to the spammer’s site), Google has announced that a rel value of “nofollow” on a link will tell their crawlers not to consider this link when calculating the link destination’s page rank. Several weblog applications have already updated their software so that links added in comments by weblog readers will automatically include this attribute setting. This gives perpetrators of referrer spam much less incentive to do add these worthless comments.

Comments added to my post of Monday (the good kind, not the spam kind) led to a discussion of how allowing people to add data and metadata to web pages that they don’t own leads to abuse, and whether the potential abuse renders user-added metadata features useless. This has always been Google’s justification for ignoring metadata, so it’s nice to see them encouraging the use of link metadata. (It was tempting to title this posting “Newsflash: Google crawlers paying attention to an attribute value besides href!”) They get extra credit for doing this with an attribute that’s been around for so long, instead of making up a new one, which is what many companies would have done.

There’s been a lot of buzz about so-called “folksonomies” lately—the use of taxonomies developed on the fly by user communities instead of by centralized developers who create a list of terms that users must choose from. Examples of the latter are used for my employer’s databases of legal material and the O’Reilly Developer Network weblogs. (What do they do with those Secondary Subjects and Topics that we select when we enter these, anyway?) The most well-known examples of the former are the tags that users can add to flickr photos and the keywords that can be added to del.icio.us bookmarks. Flickr and del.icio.us let you make up your own tags, but it pays to first query on the tags you might use to check whether others have used them and how they did so. This is what leads to the self-organizing property of the keyword collections, and it’s what gets some people so excited.

Lately, some webloggers have debated whether top-down or bottom-up taxonomy development is better, but it’s clear to me that each has big advantages. The potential for a symbiotic relationship between the two approaches within the same organization should provide fertile ground for academic research projects.

Friday’s big news in bottom-up taxonomy development was the announcement by Technorati, the weblog tracking site, that they’ll support assignment of and searching by keywords on weblog entries. It’s a big deal for web metadata because instead of creating yet another folksonomy, they’re tying together the flickr and del.icio.us taxonomies to use with their tagging system, thereby providing a model to others thinking about taxonomies of web resource keywords. It’s a big deal to people who think a lot about linking and markup because Technorati is letting people identify taxonomy terms by using the rel attribute of the HTML a element.

I mentioned above how the ability to enter a query on the use of a tag is key to the self-organization of a bottom-up taxonomy. Technorati lets us issue one query to simultaneously check for a term in the flickr collection, the del.icio.us collection, and any new weblog posts in Technorati index. I don’t even have a Technorati account, but my queries there for terms I’ve used to tag flickr pictures and del.icio.us entries turned them up the relevant pictures and entries. For example, I saw that no one had used potrzebie as a keyword on either site, so I assigned it to one of my flickr pictures and one of my del.icio.us links, and a Technorati search found both shortly afterward.

Technorati offers several methods to add tags to your weblog post so that Technorati queries find it with the flickr pictures, del.icio.us links and other Technorati posts that use the same keyword (see Technorati’s help page on this), but the most interesting method to me was the creation of an HTML link to the tag’s page using a rel value of “tag”, like this:

<a href="https://technorati.com/tag/potrzebie" rel="tag">[tagname]</a>

(As a longtime XML geek, I don’t know whether to be amused or annoyed that the new special keyword is “tag”. I try to stay amused.) I’ve written before about how little this attribute has been used, and I’ve even done a bit of research that found no use of it. As with a lot of optional metadata, people had little motivation to use this attribute because they didn’t know of applications that would do anything useful with it. Now they do: a search engine that searches by metadata assigned to images and web pages from a taxonomy developed by anyone who wants to contribute to the taxonomy.

I created a dummy sample file on Friday and manually pinged Technorati with the URL, and so far it doesn’t show up in query results for its keywords that I tagged using this markup. It has a link like the one above and another with similar href and rel values but no text between the a tags. If the latter one works, thereby letting us add metadata about our web pages to Technorati’s index without displaying anything new on the page itself, that would be really cool. I’ll add a comment below if it does.