Lists Home |
Date Index |
On Wednesday 13 February 2002 15:32, Mark Seaborne wrote:
> I hesitate to ask, but can anyone tell me what are the real differences
> between web services and the semantic web?
As I see it (and will probably now be flamed for):
Web Services are web sites providing programmeable interfaces as well as
human interfaces. This allows people to write tools that do stuff like
interact with all the auction sites out there to search them all for
something you're after. It depends on two interesting things we have yet to
1) How to find all the auction sites out there. Maintaining a list by hand
would suck. Much nicer to have a directory, like a CORBA trader, where
auction sites register saying "I provide the standard Auction Site API at
2) Standardising APIs for various things. Do you POST a request to a certain
URI and get the answer in your response? What if it's something complex and
transactional that involves much two-way communication?
The semantic web is something a bit more abstract, but more 'read only';
instead of having a Web of pages we have a Web of *information* with
rendering information attached to it. We don't need to standardise the
schemas so much as standardising the metaschemas so the browsers and spiders
can locate informational links between things; not just hyperlinks, but the
fact that one site uses a term and another site claims to offer a definition
of that term... making an implicit link that neither site needs to
'maintain'. If all references to me use some unique identifier (my urn:oid:
URN from the other thread will do :-), then documents making statements about
me can all (from software) be seen to be making interrelated statements about
the same person. This is a step above just searching on my name as a keyword,
and even a step above searching on my URI to get a list of matching pages;
the semantic web is a set of 'things' with 'relationships' between them.
Using standard relationships, for example, my URI could be related to the
string 'Alaric Blagrave Snell' with the relationship 'Is Legally Named', and
also have the property 'Is A Human Being', and related to the URI for my
company (urn:oid:1.2.818.104.22.16862548) by the relationship 'Is A Registered
Director Of', and so on. Software can know what these relationships mean;
even if the software does not know the exact relationship, it might be able
to derive (from the URI representing the relationship) metadata about the
relationship... 'Is A Registered Director Of' might declare itself as being a
more specific version of 'Is A Part Of This Group', which covers religious
affiliations, national affiliations, and so on.
(1) is more about processes, (2) is more about information.
Does that sound about right?
> Mark Seaborne
Alaric B. Snell
http://www.alaric-snell.com/ http://RFC.net/ http://www.warhead.org.uk/
Any sufficiently advanced technology can be emulated in software