Lists Home |
Date Index |
You've definitely captured what I have been trying to
say in this thread better than I have said it. Sure,
there's plenty of scope for "18th Century"
reductionist approaches to semantics, especially in
areas such as medical terminology that have been
rigorously studied since, uhh, the 18th century. I'm
not at all sure that the Web is susceptible to this
treatment (e.g., someone tell me where the xml-dev
mailing list might fit in a taxonomy of websites ... I
feel great sympathy for the people who come here
looking for straightforward answers to their XML
Also due to the fact that any ontology is the result of a social activity
and is based on a group perception agreement about things and subjects. The
whole problem of the semantic web is not to define an ontology (quite easy,
each of us can do that). The problem is to get it accepted and used by
everybody. Moreover, I am having more and more difficulties to grasp what is
the semantic web.
a) Is it a set of facts about the content published on the net
b) Is it a set of facts published on the net.
The main difference between a) and b) is simply that I can publish an RDF
document about something on the net and this document can be related to
others also published on the net. An agent can move from a document to
another and collect these facts to build a certain perception about the
world. These document are not content like we know today (i.e. HTML
documents) they are based on OWL, RDF, XTM and anything simular.
Actually nothing prevents anybody to do that. Except that it doesn't seem
that people are building a web of "view of the world" nor that these views
are connected to each other. Moreover, as we know, web pages have the
implicit notion of connection (through anchors), again, it doesn't seem to
be the case for any ontology specification where connecting ontologies
together is not explicitly mentioned nor explained (do RDF, XTM or OWL has
the notion of "links" to stitch together a web of facts about the world?).
Do anyone on this list knows a web of RDF, XTM, OWL and tutti quanti
documents that, together, create a view of the world? If yes, please don't
let us in the dark. Otherwise, let's consider that this version of the
semantic web is not there and that it is not backed by any social will.
Now about the second version: facts about the content already there. In 1995
with Guha we tried hard to create a semantic net with the MCF/MCL language.
Several "free" browsers where available to browse the web using these fact
statements (like the RDF frames of today). We are now in 2003, 8 years after
and people still do not publish
a) a sitemap about their site published in RDF or topic maps. These RDF or
XTM site maps would help an agent to get some info about the content.
b) RDF, OWL, XTM and tutti quanti statements contained in web pages.
It is not totally true to say that web documents do not include facts about
themselves. "Meta" elements are included in a lot of HTML pages. These
"meta" elements state that a page is associated to certain keywords. For
ontology inclined people, some of these keywords can be associated to a
class, an entity, an object, a subject, etc... So hold on a second, I am
wrong, we already have an embryonic semantic web!
Why search engines, the agents supposed to aggregate, process and make
inferences about these facts do not use them? Because, most of the time they
cannot be trusted. Several authors include keywords not really appropriate
to the published content (and by saying that I am very politically correct -
the reality is worse than how I describe it). in other words, agents
processing facts about the web content cannot trust these statements. How
some engines reacted to this matter of facts? By using a new way to solve
this issue. The main motto of these engines is: "if we cannot trust the meta
data accompanying the data, then we ignore it. Said differently, if we
cannot trust what they say about themselves let's listen to what the other
say about them". A linking strategy a la Google or a la Teoma appeared.
These engines are not listening to what the documents are telling us about
themselves but more to what the others are saying about them.
After nearly a decade the web is existing, we learned that meta data
provided by data provider cannot be trusted. As anybody on this list learned
after several years of existence on this planet and in our societies,
mercantile interests can corrupt any quest for truth and honesty. So, what
option is left for the semantic web then? A web of statement about the world
connected together. Instead of having HTML documents that can be interpreted
by humans, we would have a web of document stating facts about the world.
Off course, some networks would be as corrupted as our societies are.
Others, really pursuing a quest for the truth, will at least include
mechanisms preventing or controlling the excess of mercantile and political
interests. Does this web exist elsewhere than in some neuronal connections
in the brain of some individuals? Where is this semantic web? Can I browse
it? Does this "web" consist of more than one document linked together (if
not it is not a "web" it's only a singularity, a singleton). Why I do not
know the existence of this "web". Where can I get a browser for this "web"?
And finally, what's in it for me (the average question any consumer will
Actually the semantics about the web is in the hands of a few... And these
few are the search engines... This implies that the meaning of the web is
controlled by a few...
Mike you probably made the same inference as me and concluded that the main
reason we do not have yet the semantic web as a web of documents providing
facts about the world is that there are too much mercantile and political
interest roadblocks in front of this vision.
Roger you are right, any web semantics will be the result of social forces
and it is by nature holistic. And even if documents which are the result of
a reductionism approach are linked together to form a "web", social forces
will bring the whole dynamics to a different level. To really understand the
"real life" semantic web we have to study the process of how
concept/keywords/classes/subjects are associated to documents and how these
documents are ranked (given more or less importance and therefore providing
a utility value or rightness value in addition). Basically, if I think
a) I will study how the different documents providing facts or vision of the
world are linked together and what are the mechanisms that will bring
forward emergent truths through emergent "semantic webs"
b) I will examine if it is a top down (like today) controlled access or a
bottom up mechanism carried by societal networks (word of mouth or social
groups). The mechanism behind will tell me a lot about the kind of semantics
provided and if it is a "web" or simply a fantasy.
Didier PH Martin