« Good post on folksonomy; another on tagging |
Main
| Ontology as a term of art »
January 25, 2005
Why tags should be URLs
Posted by Kevin Marks
This is a drive-by response to David's 'namespaces to the rescue' post, where he says:
a table of synonyms that’s compiled manually and/or automatically by doing clustering analysis can enable us to tag local but search global. Or if generalized tag sets emerge (and I think they will, albeit not truly globally), we can use them as well as our local tags. For example, if a tag set called “AmeriTag” emerges, we could tag a photo as [ameritag:hotdog food_eating_contest obscene_idiots], where the second tags are purely our own. (Namespaces to the rescue!)
This is exactly why the
rel="tag" model takes a URL (not a URI) as the tag parameter. URLs have namespaces. In this case we have a choice of tagspaces - using http://technorati.com/tag/[tagname] is one way to do it, but of course you can use Wikipedia or AmeriTag or whichever disambiguated tagspace you like as your link.
However, by making them URLs that actually locate something useful, they make sense as visible links within your posts. (more technical details here)
Comments (11)
+ TrackBacks (0) | Category:
- RELATED ENTRIES
- › Spolsky on Blog Comments: Scale matters
- › "The internet's output is data, but its product is freedom"
- › Andrew Keen: Rescuing 'Luddite' from the Luddites
- › knowledge access as a public good
- › viewing American class divisions through Facebook and MySpace
- › Gorman, redux: The Siren Song of the Internet
- › Mis-understanding Fred Wilson's 'Age and Entrepreneurship' argument
- › The Future Belongs to Those Who Take The Present For Granted: A return to Fred Wilson's "age question"
1. orcmid on January 25, 2005 1:26 AM writes...
If the XML notion of namespace is meant here, the names of namespaces are indeed URIs, not URLs and there is no requirement that there be anything at the target of a URL used for such a URL, intelligible or otherwise. Sorry. It's an out-of-band deal with the namespace URI as a discriminator of namespace identity. Heh.
But I'm not sure that is material to Clay's observation and I will desist. I love the earlier rant about ontologies. I think the problem of constraining the future is exactly where it comes apart, along with misguided and simplistic forcing of conceptual hiearchies at the outset.
Permalink to Comment2. orcmid on January 25, 2005 1:32 AM writes...
Oops. OK, I also don't think URLs have namespaces at all (there's no one-to-one about namespaces), and I apologize for blurring Kevin and Clay together here. And I strongly support Clay's view with regard to ontologies.
Permalink to Comment3. Peter Clay on January 25, 2005 3:56 AM writes...
What do you do when the URL breaks? Is the tag declared dead? Do you have "semantic rot"?
Permalink to Comment4. James Tauber on January 25, 2005 8:01 AM writes...
How about Wikipedia URIs as suggested in Wikipedia as a URI Lookup Service
Permalink to Comment5. joe on January 25, 2005 12:17 PM writes...
I use Wikipedia all the time. Kind of like I use networksolutions.com.
Permalink to Comment6. Kevin Marks on January 25, 2005 2:36 PM writes...
orcmid:
Permalink to CommentThe point is that an abstract namespace is unnecessary if you have a concrete linkable tagspace.
James:
yes, wikipedia is a fine tagspace, and I had mentioned it in my posts discussing this.
Peter:
if the URLs break, they are no worse than namepsaces start off with.
7. Tim on January 25, 2005 4:19 PM writes...
It's interesting to see the different methods of tagging popping up.
SocialSoftware
social software
social_software
social+software
They're personal ways of establishing further control or hierarchy to the tagging system. It works well for us individually in the short term, for example, to keep apple_computer separated from apple_pie_recipe, but in the long term it corrupts the data for others in the same way as ontology does.
We need to trust that decent search tools will come along to help us filter words with multiple meanings. We trust Google, but not Delicious. How hard would it be for Google to spider a page ifallofourwordsrantogether?
Permalink to Comment8. kevin Wen on January 25, 2005 6:52 PM writes...
hi, Kevin, i think this is worthy to check out.
http://www.technorati.com/tag/10placesofmycity/
"10 Places of My City" is Chinese blogger social movement by using Technorati tag system to encourage bloggers to showcase the top 10 places of their own city .
Over 40 Blogger join this, post their favorite places, spreading the ideas.
Permalink to Comment9. Elias Torres on January 25, 2005 8:27 PM writes...
RDF does this out of the box. I just hope more people look into RDF before reinventing it all over again. Although, changes like this will make it much easier to transform existing web content into RDF metadata.
http://torrez.us dc:category technorati:family
http://torrez.us dc:category technorati:photos
http://torrez.us dc:category flickr:kids
BTW, it does not have to be dc:category.
Anyways, I agree with Kevin that tags should be URIs.
Permalink to Comment10. kael on January 28, 2005 6:20 PM writes...
What about The 'tag' URI scheme ? http://ietf.org/internet-drafts/draft-kindberg-tag-uri-06.txt
Permalink to Comment11. Kevin Marks on January 28, 2005 7:03 PM writes...
Elias: I said URLs, not URIs - I want them to be resolvable.
Permalink to Commentkael: The tag: uri scheme is something else - it is a way to construct GUIDs without reference to DNS, and is not resolvable. If you put it in a a href= link it will give an error page when clicked on.
I want to avoid the 'RSS icon' aversion conditioning, where the reader clicks on one once, gets a page of gibberish, and resolves never to do that again.