W3C home > Mailing lists > Public > www-rdf-interest@w3.org > January 2001

Gimme some ideas

From: Joshua Allen <joshuaa@microsoft.com>
Date: Mon, 22 Jan 2001 00:58:29 -0800
Message-ID: <4F4182C71C1FDD4BA0937A7EB7B8B4C10FEF8B@red-msg-08.redmond.corp.microsoft.com>
To: <www-rdf-interest@w3.org>
Ok, I know this is probably boring for you smart people, but I have been
considering the practicality of various uses for RDF in enabling the
"semantic web".  I'm sure that many of you have thought deeply about
this, and I am hoping you could fill in some gaps I have, offer
opinions, etc.  So anyway, here are my notes I would start from if
attempting to explain RDF to my grandmother; areas of particular
uncertainty marked as such..
 
- currently our computers are good at processing words, documents,
sentences in ways that do not depend on knowledge of what those words
mean.  Bold formatting, text alignment, etc. are all things we do with
words in a meaning-agnostic manner.
- semantics is "what a word means".  semantics could be as simple as,
"the word Microsoft is usually a proper noun that names a company", or
it could be as complex as the complete body of human knowledge about
"Microsoft"
- computer systems that understand some degree of semantic information
about the document being processed can do various "smart things" with
that data.  For example, when I type the sentence "Yesterday IBM
announced earnings" in Word 10, a feature called "smart tags" recognizes
and faintly marks "IBM".  When I hover my mouse over the word "IBM", I
can display a popup menu that lets me get stock quotes for IBM, current
news about IBM, and so on.  Netscape 6 allows smart behavior based on
semantic information stored about the page you are currently viewing.
- computers can get semantic information about a particular text in a
number of ways.  I can think of three big ones: markup embedded directly
in the text, semantic info about the text stored in some external
database, and semantic information derived "on-the-fly" through analysis
of the text using grammar parsers, lexicons, etc.  (Of course, the first
two options could be accomplished through use of some AI as described by
the third.)
- <<??>> RDF is mostly about the first option.  If I'm storing RDF in a
database, chances are I am going to have a nice API that I can use to
get the semantics the way I want them, right?  The reason I use RDF is
because I want to provide semantic information *with* a page (yeah,
resource, whatever).  I'm not saying the RDF has to be *in* the page,
just shipped *with* the page.  [So is this what people are thinking?
Web pages marked up (associated) with RDF all around the web?  I'm in no
way saying that is the *only* use of RDF, but that sure seems like the
most promising "big step" toward a semantic web...]
- <<??>> I could use RDF to store semantic info about a page, or I could
use it to mark up phrases and words *within* a page.  [Do people have a
preference one way or another for this?  Is there something
fundamentally screwed up about having blocks of text scattered
throughout a page all marked up with different RDF characteristics?]
- Trust and "authoritative-ness" are certainly issues for the future,
but there are plenty of cool semantic things that can be done without
working out those issues yet.
- So far, this seems like a good target -- if applications do smart
things with semantic information, users will have incentive to provide
semantic information with their documents.  And if semantic information
is widely available with texts, applications will have a bigger
incentive to use that information.
- RDF is about interoperability of semantic markup.  Needless to say,
not all semantic information that applications store and use to do
"smart things" cries out to be widely publicly consumable.  RDF is about
that semantic information that is most useful shared, and provides a
standard so we can amplify the effect cited above.
- <<??>> On the other hand, a standard always needs a big user community
to legitimize and protect its existence, and a strong user community is
usually the result of a killer app or two.  The killer apps are going to
sway the way that the standard is used in the future.  [What are the
ways that you do NOT want to see semantic markup used in the future, and
why?]
 
 
Received on Monday, 22 January 2001 04:00:33 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:47 GMT