So the 'usual story' for an apple would include growing on a
tree in an orchard, or in the wild, and being picked by a
human, transported to storage, offered for sale, bought,
cooked, eaten, excreted.
Arranging the k-base as story-skeletons has several advantages
over Cyc-- eg the most common consequences of any event are
stored in sequence and can be retrieved effortlessly, and
the human knowledgebase-builders can spot gaps more easily
because the story-format is intuitive to them.
The mailinglist limped along for a year or so, but no one
ever really undertook to start cranking these out. (The
archives were at my old ftp-site until recently-- I'll
restore them sometime, probably.)
But I now see a way of moving forward that's much less
quixotic/herculean, via distributed Web-authoring.
Over the last five years I've been evolving a web-design
style that merges a usual-story-like 'faq' with an
exhaustive survey of Web-resources on the topic. Almost
always this can be focused around a timeline, with Web-
links sorted into the timeline, often several links for
each timeline entry.
These pages aim for the most concentrated _utility_ I can
conceive, and so their shape has a sort of 'necessity' in
comparison to the arbitrary 'contingency' of most webpages,
which rarely try to be faqs, and almost never merge the
links with the faq-text.
So my vision now is of "the Necessary Web" which will
someday offer one or more 'necessary'-style pages on every
important topic, linked into an increasingly exhaustive
encyclopedia of the world and the Web. I'm hoping that
some enlightened universities will recognise this as an
ideal activity for students as they explore almost any
subject, and that different universities might even compete
to offer the highest-quality alternative editions.
I believe this is a much more realistic approach to the
Semantic Web than via XML, because it starts with the
hard issues-- the semantics-- rather than rolling out
an expensive syntactic framework (XML) with no agreement
on the semantics required to make it 'go'...
My necessary-style pages naturally sort their content
into semantic clusters, some of which are unexpected.
Eg any page about an artist/author should include a
section on the other artists who inspired him, and this
may be further sorted chronologically-- early influences,
later influences. (Cf topic maps in XML)
Discovering these necessary semantic shapes ought to be
an exciting process, and the lessons learned in building
each small piece should make later pieces fall into
place more easily.
I can imagine the basic backbone coming together very
quickly, and at that point some real XML-style semantic
markup should become practical, with the markup embedded
in the necessary-pages effectively providing a Cyc-kernel
that will enable smarter and smarter artifical
> Over the last five years I've been evolving a web-design
> style that merges a usual-story-like 'faq' with an
> exhaustive survey of Web-resources on the topic. Almost
> always this can be focused around a timeline, with Web-
> links sorted into the timeline, often several links for
> each timeline entry.
> eg: http://www.robotwisdom.com/jorn/pullman.html
I saw this style with the "early jazz" page you directed me to in another
thread. This is a simple, but great, idea that I think I am going to put to
use with some of my students this semester. I agree that starting with the
framework makes things unnecessarily complicated. I also find that
students, when they are working on "web-quest" kinds of ideas tend to not
only separate the information they are presenting from their links in a way
that decontextualizes the linked material (to their detriment), but they
get caught up in design dictating their structure, which always hurts them.
From the information, a practical framework can be derived.