W3C Issues RDF and OWL Recommendations

[Andy Dingley]

Semantic Web emerges as commercial-grade infrastructure for sharing
data on the Web

http://www.w3c.org/2004/01/sws-pressrelease
New docs are up - congrats to Brian and everyone else involved. Now
lets get out there and build something with them !

[Spartanicus]

Andy Dingley <di*****@codesmiths.com> wrote:

Semantic Web emerges as commercial-grade infrastructure for sharing
data on the Web

http://www.w3c.org/2004/01/sws-pressrelease

I briefly glanced at w3c's documents on the new meta data schemes such
as Dublin Core a few months back. I could not see the point, author
defined meta data in the form of <meta> tags has failed, afaics not
because of a lack of definition, but because the principle of author
defined meta data is afaik fundamentally flawed.

Proper HTML markup only offers very basic information about semantics,
thus placing the task of indexing data with the indexing mechanisms, but
it seems to me that it's the only mechanism that has some chance of
limiting abuse by content authors.

Am I missing something? If so, could someone define the problem these
new meta data schemes are supposed to solve?

--
Spartanicus

[Tim]

Andy Dingley <di*****@codesmiths.com> wrote:

Semantic Web emerges as commercial-grade infrastructure for sharing
data on the Web

http://www.w3c.org/2004/01/sws-pressrelease

Spartanicus <[e-mail]> wrote:
I briefly glanced at w3c's documents on the new meta data schemes such
as Dublin Core a few months back. I could not see the point, author
defined meta data in the form of <meta> tags has failed, afaics not
because of a lack of definition, but because the principle of author
defined meta data is afaik fundamentally flawed.

Proper HTML markup only offers very basic information about semantics,
thus placing the task of indexing data with the indexing mechanisms, but
it seems to me that it's the only mechanism that has some chance of
limiting abuse by content authors.

Am I missing something? If so, could someone define the problem these
new meta data schemes are supposed to solve?

They look more like something of interest to someone designing a
database, and wanting to do it in a standardised, expandable, and
defined, manner (e.g. museums, libraries, researchers, etc.). For
people who're trying to do the right thing. Anybody setting out to just
exploit a system is going to do that, anyway (any way that they can, and
regardless of consequences).

The current search engine cataloguing process just seems to take a look
at all the viewable HTML content, and assess the lot for potential
indexing (almost like scanning a plain text document). Sure there's
some criteria to rate importance (headings, titles, etc.), but there's
little other ways to properly catalogue the data. And weeding out
results in the search engine is a hit an miss affair, mostly based on
keywords, rather than semantic descriptors (e.g. I want a datasheet on
some 555 timer IC, I can't automatically and positively exclude things
that aren't datasheets, because there is no such classifier; I can only
hope that something mentioning a datasheet is one, rather than something
that just mentions the word).

That's an over-simplified example, I know, but discusses the general
gist of it. There's no proper process for doing a specific search, it's
(nearly) all keyword based.

--
My "from" address is totally fake. The reply-to address is real, but
may be only temporary. Reply to usenet postings in the same place as
you read the message you're replying to.

This message was sent without a virus, please delete some files yourself.

[Karl Smith]

Andy Dingley <di*****@codesmiths.com> wrote:

Semantic Web emerges as commercial-grade infrastructure for sharing
data on the Web

Great!
1. Where do I download the free browser?
2. What are the URLs of some commercial-grade websites using this new
"infrastructure"?

[Karl Smith]

Spartanicus <[e-mail]> wrote:

Andy Dingley <di*****@codesmiths.com> wrote:

Semantic Web emerges as commercial-grade infrastructure for sharing
data on the Web

http://www.w3c.org/2004/01/sws-pressrelease
I briefly glanced at w3c's documents on the new meta data schemes such
as Dublin Core a few months back...

[DC has been around for years, is successful within its community of
interest, and has nothing to do with W3C.]
Am I missing something? If so, could someone define the problem these
new meta data schemes are supposed to solve?

OWL is not about metadata, but actual data. Rather than writing
webpages for humans to read, we're supposed to write "ontologies" for
bots to read. We all spend a day/week/year of our lives putting
everything we know about some subject into an ontology, then plonk it
on a webserver somewhere and pass the URL to the U-beaut new OWLbot.

The OWLbot crawls everyone's ontologies, reconciles them into a
super-ontology, and "presto!" it knows everything about everything.
Then, if you are a good boy, you can query the super-ontology to find
out what you want to know. It's the old "General Problem Solver" scam
that MIT pulled back in the 1960s re-invented under a new name.

It will (probably) fail because it is based on the false premise that
I, you, and everyone else owe the W3C/MIT/USgov a favour and want to
surrender any subject-specific expertise we might have into their
hands and thus render ourselves redundant.

--
<?UniversalTranslator
input="Hov ghajbe'bogh ram rur pegh ghajbe'bogh jaj"
lang="x-klingon" ?>

[Jeen Broekstra]

Karl Smith wrote:

Andy Dingley <di*****@codesmiths.com> wrote:

Semantic Web emerges as commercial-grade infrastructure for
sharing data on the Web
Great!
1. Where do I download the free browser?

If RDF-enhanced browsing is your interest, you might want to
check out Magpie (don't have the URL handy, sorry). It's a
browser plugin that enhances web pages with annotations based on
a pre-selected ontology. Nifty stuff.
2. What are the URLs of some commercial-grade websites using
this new "infrastructure"?

Most of the commercial applications I have seen sofar mainly
target corporate intranets and as such are usually not visible
from the outside.

But the FOAF project (http://www.foaf-project.org/) is an
interesting starting point if you want to learn more about what
this stuff can do.

Jeen
--
Jeen Broekstra http://www.cs.vu.nl/~jbroeks/

When a lion meets another with a louder roar,
the first lion thinks the last a bore.
-- G.B. Shaw

[Michael Bauser]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Spartanicus <[e-mail]> wrote in message
news:<2r********************************@news.spar tanicus.utvint
ernet.ie>...

I briefly glanced at w3c's documents on the new meta data schemes such as Dublin Core a few months back. I could not see the point, author defined meta data in the form of <meta> tags has failed, afaics not because of a lack of definition, but because the principle of author defined meta data is afaik fundamentally flawed.

You missed the *biggest* difference between HTML meta elements
and RDF -- RDF isn't limited to author-generated assertions.

That is: HTML meta elements can only refer to the page they're
part of, so there's no way for *me* to create HTML metadata
about *your* pages. RDF, on the other hand, lets anyone make an
assertion about about any resource, using a format that makes
comparing descriptions easy. So, you can create RDF that says
"I've got the greatest homepage ever", and I can create
compatible RDF that says "No he doesn't", and the end-user can
read one, the other, or both.

Metadata in an HTML universe is a question of "Do I trust the
author?"; metadata in an RDF universe is a question of "Who do I
trust more?" After that, it's all about competition, reputation,
and trust. If you earn a reputation as more trustworthy than me,
than people trust your description of your homepage. If I'm
considered more trustworthy, then people trust my description of
your homepage.

And I know somebody out there is saying "but that's too
complicated". It shouldn't be -- it's the decision-making
process people use in the real world. For example: Which
description of a book do you trust more -- the publisher's
description on the back cover, the professional book reviewer's,
your best friend's, or the library card catalog's? That's a
decision about trustworthiness that you're making based on your
own experiences. RDF metadata can work the same way *if* there's
enough produced, *if* it's attached to people/groups that can
earn reputations (completely anonymous metadata is not so
useful), and *if* user agents are created that let end-users
manage their assessments of trust.

The RDF and OWL standards are Step One of Three. Now, people
have to start producing the metadata, and programmers have to
start making software that will use it. (Like many W3C
initiatives, it's going to take a while.)

In some ways, RDF is a format waiting for its killer app to show
up. It might not be "web descriptions", it might be "people
descriptions" (like the FOAF project), or it might be something
none of us have thought of yet. Still, there are concepts there
that are too useful to dismiss out of hand.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (MingW32) - GPGshell v3.01

iD8DBQFALJGVcpuEJT2bpHsRArv5AKDJiK6snP6uFOq/lL1whKPcDP5XEgCgkac+
gwZaCRHEv/YeE+tJ8Gb3N5Q=
=0sIJ
-----END PGP SIGNATURE-----

[Jim Ley]

On 12 Feb 2004 06:16:39 -0800, go************@kjsmith.com (Karl Smith)
wrote:

Andy Dingley <di*****@codesmiths.com> wrote:

Semantic Web emerges as commercial-grade infrastructure for sharing
data on the Web

Great!
1. Where do I download the free browser?
2. What are the URLs of some commercial-grade websites using this new
"infrastructure"?

news.bbc.co.uk (RSS)
livejournal.com (FOAF)

Jim.
--
comp.lang.javascript FAQ - http://jibbering.com/faq/

[Tim]

Andy Dingley <di*****@codesmiths.com> wrote:

Semantic Web emerges as commercial-grade infrastructure for sharing
data on the Web

go************@kjsmith.com (Karl Smith) wrote:
Great!
1. Where do I download the free browser?
Why does it have to be free? Sure, it's nice when someone releases
their work for the good of all, but there's no onus for people to work
for nothing.

2. What are the URLs of some commercial-grade websites using this new
"infrastructure"?

1. Standards have to be created before people can use them. Have
patience.

2. Not all publicly available data is published for everyone in the
public to use. e.g. Your laws are publicly published, but only those
interested will read them. Defining how to do something is useful for
those using that information, at the very least. If the scheme is
useful, and expandable, it might be applied to other things.

Some people just like to go around shooting their mouth off, it seems.

--
My "from" address is totally fake. The reply-to address is real, but
may be only temporary. Reply to usenet postings in the same place as
you read the message you're replying to.

This message was sent without a virus, please delete some files yourself.

[Reinier Post]

Karl Smith wrote:

It will (probably) fail because it is based on the false premise that
I, you, and everyone else owe the W3C/MIT/USgov a favour and want to
surrender any subject-specific expertise we might have into their
hands and thus render ourselves redundant.

Not only that, even if you and I can be paid indefinite amounts of money
to do this as unbiasedly and hifily as we can for an unlimited amount
of time, full-text searching will simply provide better results.
This was known in the 60s; considering the relative development
of quality/cost ratios for manpower and computer power, I see little
reason to revise that assessment. Google and its like prove it, by the way.

--
Reinier

[Tim]

Karl Smith wrote:

It will (probably) fail because it is based on the false premise that
I, you, and everyone else owe the W3C/MIT/USgov a favour and want to
surrender any subject-specific expertise we might have into their
hands and thus render ourselves redundant.

That seems to be making a rather large jump to a conclusion, based on a
very narrow viewpoint. If providing good detail is writing yourself out
of a job, then providing any information is the same.

If I was publishing photos, for instance, for the public, and wanted
anybody to be able to find something in particular by certain criteria,
then it would be in my interest to properly classify my resources. If
my market is over the internet, then making them easily findable over
the internet, is definitely to my advantage.

On the other hand, if my job was fixing PCs, then it wouldn't be
sensible to publish anything about how to do it yourself, never mind
making it easy for a search engine to find it.
[e-mail] (Reinier Post) wrote:
Not only that, even if you and I can be paid indefinite amounts of money
to do this as unbiasedly and hifily as we can for an unlimited amount
of time, full-text searching will simply provide better results.
This was known in the 60s; considering the relative development
of quality/cost ratios for manpower and computer power, I see little
reason to revise that assessment. Google and its like prove it, by the way.

Which would be due to the data being collected, and the manner that it's
collated. Given properly described resources, and suitable capable
search engines, a properly classified system would be better. Google
proves nothing, because you don't have the opposing idea to compare
with.

Try using a properly classified system, like a well designed library
catalogue, and you soon find out it's superior. e.g. If I wanted to
research books by a certain author on certain subjects, I can. The
search engine will ignore unrelated mentions of the keywords (such as
the author's name in some other book's title).

--
My "from" address is totally fake. The reply-to address is real, but
may be only temporary. Reply to usenet postings in the same place as
you read the message you're replying to.

This message was sent without a virus, please delete some files yourself.