Tid for ømhet
Det hjelper lite med kurs og kompetanse hvis vi ikke har tid. Juristen skriver dårlig, vi sender ham på skrivekurs. Saksbehandlerne eier ikke språkøre, vi arrangerer et kurs i nettskriving. [...]
Hot on IA Summit: Semantic web and permanent links 4
Chris Thorne from BBC held an interesting speech on the IA Summit regarding the importance of persistent URIs in building a semantic web.
What is the semantic web?
The Internet is designed to be understood by humans, not computers. When reading the word “Apple”, the first thing that comes to mind may be a fruit, an electronics and software company or a record label. Computers have a hard time understanding the difference. The semantic web is a web that machines can understand, by publishing in a language that computers can process and link.
Why is persistent URIs important in a semantic web?
Prince Rogers Nelson, better known as Prince, has been known by various names through his career, amongst them simply a symbol:
Between 1993 and 2000, he was referred to as The Artist Formerly Known as Prince. The problem then is, that if the URIs of pages about Prince should change every time he rebranded himself, all links to those pages would break, and the Google PageRank would drop (since fewer pages are linking to them). BBC is collaborating with Musicbrainz, which has developed a system of unique identifiers for all artists, releases, tracks, labels and discs. This allows for permanent links to artists, even though they change their name. The unique ID for Prince on Musicbrainz.org is 070d193a-845c-479f-980e-bef15710653e, and by using the same identifier, several organisations are able to identify and exchange data about this artist.
http://www.bbc.co.uk/music/artists/070d193a-845c-479f-980e-bef15710653e
http://musicbrainz.org/artist/070d193a-845c-479f-980e-bef15710653e.html
http://www.last.fm/mbid/070d193a-845c-479f-980e-bef15710653e
How user friendly is this approach?
http://www.last.fm/mbid/070d193a-845c-479f-980e-bef15710653e is ugly. It’s impossible to deduce from the URI what page to expect, let alone remembering the address. So, is the use of unique identifiers a step backwards in terms of usability? Last.fm’s approach is to forward the user to http://www.last.fm/music/Prince, which undoubtedly is easier on the eye. However, forwarding is but a temporary solution, and it is said to have a negative effect on the Google PageRank.
Cool URIs don’t change, Tim Berners-Lee
Do we have to make a choice between readability and a semantic web, between humans and computers? Could we envision a web with a dual address system, where semantic links are used and understood by computers only, while readable versions are used and shown to humans?
See also Chris Thorne’s presentation on Slideshare.
Andreas er stor tilhenger av gjøre ting enklere. Oppgaver kan alltid bli enklere og mer intuitive. Han mener at nettsteder bør lære av og tilpasse seg brukerne, ikke omvendt. Spesielt på mobiltelefoner er det å gjøre det enkelt ekstremt viktig for å lykkes.
Vegard Sandvold, 04.04.2009 14:09
Redirecting from one URL to another will damage your Google PageRank if you’re not careful. Using a “HTTP 301 Moved Permanently” is recommended. You can find several examples on this page: http://www.webconfs.com/how-to-redirect-a-webpage.php
People will always use different names when referring to a specific subject. So it seems very artificial to force human readable names and labels to be unique. It’s better, as you’re suggesting, to have machines translate names/labels to unique identifiers.
You can find more opinions about identifiers for semantic web in the comments section here – Wikipedia – A Democratic Gold Standard for Topic Maps
Svein Ølnes, 12.04.2009 13:50
Seems like a bad idea indeed. We already have two standards for describing unique resources like this; W3C’s RDF/OWL and the ISO standard Topic Maps. We should not try to develop yet another standard unless there are significantly shortcomings to the two standards previously mentioned.
In Topic Maps a resource, or rather a topic, is given a unique identification using a PSI -a Published Subject Identifier. A PSI is expressed as a URI and it is higly recommended, if not a requirement, that these are expressed in a human-friendly way. In addition to the unique URI the PSI should also have a description of the resource/topic.
Steve Pepper’s paper The Case for Published Subjects (highly recommended) suggests these requirements for a global identification scheme:
1. The mechanism as a whole should be
– democratic
– scalable
– easy to adopt
2. The identifiers themselves should be
– easy for humans to use
– easy for computers to use
Note that readability for humans is put before machine-readable.
Pingback: → Twitter Trackbacks for Hot on IA Summit: Semantic web and permanent links | IAllenkelhet [iallenkelhet.no] on Topsy.com
Pingback: → UX-konferanser første halvår 2010 | IAllenkelhet