NZ: Remote Tagging and Reverse Relationships

In Remote Tagging: A Richer Social Model, Stowe Boyd describes an addition to the RelTag microformat that allows shared tags themselves to be embedded in the link. He describes this as remote tagging:

This type of anchor operates directly on the URL it is referring to, adding a tag based category in addition to the basic link itself.

This would also provide tremendous fodder for analysis of the social networks implicit in links. For example, if I have 10 links to Marc's Voice, and 7 are tagged "socialarchitecture" and 3 are tagged "deathtopanelsessions", the nature of our social involvement can be teased out. And likewise, the multifaceted nature of people's social networks could be directly supported in this way. I could tag all links, including blogroll entries, so that the various overlapping social networks that comprise my world could be evident. This would mean that we could drop efforts like FOAF, and instead simply enrich the blogging activities we already are involved in.

If you change the phrase "blogging activities" to "publishing activities", you will understand why this is such a timely consideration for me. I have recently started working on a proposal for a New Zealand community metadata standard, and immediately ran into some of the semantic issues surrounding the way that the rel attribute is used in HTML.

I'm starting to think of the remote tagging concept as something like "attribute overloading", for the rel tag, and I think there are two distinct approaches:

The first is the "operator namespace" concept, demonstrated where the relationship type is used as a namespace for an actual category. This is the approach outlined by Stowe's approach to link tagging, where the rel attribute contains the information about what the type of relationship is, and a namespace term as an instance of that type, so tag:design, topic:architecture, etc.

The second is the "property namespace" concept, which treats the relationship as a property of a parent type or vocabulary, like nz.topic, nz.subject, etc.

If possible - I would like to establish a standard for adding richer relationships to links without needing attribute overloading in HTML at all. The main reason for this is to ensure simplicity. The attribute syntax in HTML and XML serves a specific semantic purpose, as does the rel attribute itself, and moving further in this direction defeats one of the basic purposes of XHTML, which is to allow these kinds of modular extensions to the language. As far as possible, we should use both HTML and XHTML for what they were designed for.

It's also clear that there are two kinds of relationships evident in this rich linking between sites, and this is summed up in the basic design of HTML. The cannonical standard for link relationships is the rel="index" statement, which basically says: "This link goes to the index page for the current page". Which is why rel="tag" works for Technorati, as this relationship states "This link goes to a tag page for the current page".

Remote tagging could be a reverse relationship. The statement rel="tag:architecture" means "This link goes to the architecture tag page for the current page", when it seems that a lot of the time the author may actually want to mean: "This page is adding the architecture tag for this link".

My proposal for this form of remote tagging is thus:

This denotes a specific kind of link, with special consideration placed on the title attribute. Not every link should be marked up like this - it should be reserved for considered and specific usage, as it could be a powerful strategic tool if adopted in a constructive fashion - it could also be an easy way to make a great deal of mess!

What I am most interested in is how leveraging the specific uses of rel and rev in unison could open up the potential for these relationships being useful to a broader audience of web publishers, moving away from the blog-centric view that seems to be driving this technology at present.

The New Zealand focused idea could work, because it represents a small, yet totally diverse group with a very strong awareness of identity. Projects such as Te Ara and Matapihi are just the beginning, showing how centralized metadata can be used to make a vast archive of institutional resources available to a wider community. But decentralized metadata is equally important - my idea is that a standard must exist to support a grassroots framework that brings the next generation of distributed web services to New Zealand content with as low barrier of entry as possible. It doesn't have a name yet, but I'm working on it...