Re: Semantic Web gTLDs

--- muguet@ensta.fr wrote:



[...]



> One may ask questions why the SW
progress has been so slow.

> If there is any studies on this, please let
me know.



Let me venture a guess: Creating metadata costs effort, spending
this effort is hard to justify if there are no applications that show the
benefit of creating the metadata.



Creating Semantic Web applications is
difficult if you don't have data, it makes it hard to show that these applications
are cool. 



These two things work together to create a massive cold start
problem; a new tld is not going to do anything about that. 



>

> Use of
metadata has been reduced in comparison to the early days of

> the Web. Except
for specialized ones, search engines are not relying on

> Metadata anymore,
because metadata have been spoilt by pornographers

> and "referencing" companies
that are using any conceivable trick

> to cheat the referencing algorithm
of search engines that must

> constantly evolve to fight those tricks. There
is a secret war

> going out there.



SWepam (Semantic WEb sPAM?) has been
talked about a lot, but imho it still has to materialize on a big scale and
some metadata standards managed to evolve by just ignoring this issue (RSS
or the relTag microformat). Even when you start to worry about trust, there
got to be less centralized-one-size-fits-all approaches.  



>

> One goal
of a SWgTLD is a get clean, unspoilt, METADATA.

> Sites that are purposedly
spoiling their metadata, are removed

> from the SWgTLD registry.



Who's
going to decide what "spoiled" metadata is? If it is you, I would image we
would never see "unspoiled" pornography metadata accepted into the SWgTLD
(even though you could accurately describe the content of a porn site). The
more expressive the metadata gets, the more problems you run into: what about
a site that advertises its Vitamins as "Treatment for Cancer"? It may not
be supported by medical research, but there is a large minority that things
this is true and would consider this metadata helpfull. And if you really
start to put ontologies on the web: what about the site that classifies someone
like Kissinger as murderer? Gives the age of the world as 6000 years? 





cu



Valentin Zacharias

FZI (Research Center for Information Technologies)

http://www.fzi.de/ipe

Blog: http://vzach.blogspot.com/







p.s. please
excuse the strange email address (created by a news reader that aggregates
information from different rss files and mailinglists). My real address is
zach@fzi in Germany (tld="de").

Received on Saturday, 8 October 2005 04:43:13 UTC