[Date Prev] [Date Next] [Thread Prev] [Thread Next] Indexes: Main | Date | Thread | Author

Re: [ba-ohs-talk] Node Sequencing [Was: **** Instant Outlining !!! ***]

Alex Shapiro wrote:    (01)

>> The big
>> clinker is not so much the ability to sew all those nodes together
>> in a big, multi-threaded, multi-dimensional representation (heck,
>> that's *hypertext*), it's taking *that* and adding the *other* view
>> of the information obtained via a topic-based view, ie., an ontology-
>> informed view. This is why I named my language (when I looked it up
>> again yesterday) the "XML Sequence Map Markup Language (XSQM)"; I
>> think the thing missing has been the organizing principle behind the
>> links, which is informed by the ontological content. Otherwise, it's
>> just a big ball of string. I wanna see the overlap of those two maps.
>> *That's* where I'm heading, and why I'm excited.
>> Murray    (02)

 >    (03)

> :) :) :)
> Ok, first of all, let's make it "That's where *we're* heading", and I am 
> not just talking about myself here.  My opinion is that whether we like 
> it or not, we (people in general) are going to have to more in ?this? 
> direction because it represents the solution to several problems that we 
> are currently grappling with.    (04)

Well, I can only speak for myself, hence my choice of words. I long
ago stopped expecting people to understand me or agree with me, now
it's simply a matter of hope. But yes, I certainly agree that if we
are to solve some fundamental knowledge representation problems, we
first require a means of capturing knowledge in potentially as many
dimensions as mirrors the reality we're trying to represent. Now, I
recently had a discussion with a string physicist (!!) who made the
claim that their (i.e., the consensus among string physicists) best
description of physical reality is somewhere in the neighborhood of
eleven dimensions. I really don't expect that my ontologies will be
using eleven dimensions-- the point is, any representation notation
system should be flexible enough to handle that. What if there were
a desire to use an ontology to represent string theories? That sure
seems like a reasonable idea.    (05)

> Now, I could try to start naming those problems, and specifying which 
> direction it is that I am talking about, but I don't have the words for 
> that right now (or too many words), so instead I think that I'll head in 
> the direction of greatest information density.    (06)

I'd start with the simplest, but perhaps we can meet in the middle.    (07)

> Above you say that you want to see the overlap of those two maps.  I 
> think that you are actually talking about three maps:
> 1. Ontologies = faceted classification.  http://www.aidministrator.nl/ 
> has had the effect of equating the two terms for me.
> 2. Graphs + hierarchies.  On second thought, Graphs is really what we 
> are talking about here since hierarchies are subsumed by faceted 
> classification. (As implemented at 
> http://bailando.sims.berkeley.edu/flamenco-interface.html ) Graphs are 
> not subsumed by hierarchies since they can have unnamed edges.  
> Otherwise, each edge could have a corresponding facet.  Even so, having 
> facets with only two members is not practical for current 
> implementations of FC, while it is for graphs (it being having some 
> unique type of edge that does not appear elsewhere).    (08)

I don't see the differentiation between the various things you
mention above to see them as distinctly requiring a different
category. It certainly seems possible (given the plethora of
available options, with new ones appearing almost daily) to
use a graph formalism to represent ontologies. One can use
graphs (ie., typed relations within graphs) to represent
hierarchies, as evidenced by say, the GZigZag/Gzz usage
of the same. See [1] below, a reference to the Gzz doc
describing xanalogical structures, which from what I
can see entirely graphs, or representable as such.    (09)

> 3. Time dependent data, order, and implication (Stay with me here, I 
> hope that my explanation for #2 did not put you to sleep). *Trails* is 
> another word for it.    (010)

I'd been calling them timelines and following some discussions
with Eliot Kimber had written up a rather weak timeline markup
language which I abandoned once I became acquainted with topic
maps, seeing them capable of doing the job with more aplomb...    (011)

> Right?  That's what we are talking about here:  Sequence and 
> Association, Left brain and right brain 
> http://www.viewzone.com/bicam.html (see table in the middle of page, 
> maybe I'm being flaky about this brain analogy).    (012)

You shouldn't go there. That's turned out to be a scientific
cul de sac, from what I understand. But I get the point, and
this is what I was mentioning yesterday: that not only do we
need to describe reality in some fixed way (since it's not a
static thing), we need *temporality*. I don't think it any
coincidence that some of the brains behind Hytime are also
some of the brains behind topic maps. Like Doug Englebart,
these guys were there a few years before us (I always feel
lately like I'm playing catch-up, though with the exception
of Eliot I'm younger than all of them... :-)).    (013)

> Ok, it would seem that we are back to two things (to me as well), but I 
> just realized how it's still three.
> 1. Hierarchy   (faceted classification)
> 2. Association (graphs )
> 3. Sequence    (blogs )
> We are trying to come up with a system that allows us to handle all 
> three.  And it is possible.  Where there is a clearly stated problem, 
> there is a solution (or a proof for why a solution can not exist).    (014)

Well, I'll stick with my guns and say that #1 can be represented by
#2. In fact, the XTM topic map syntax as delivered includes a set of
PSIs that includes superclass and subclass, which is all you need for
heirarchy, well, all of the essentials anyway, and any relation types
missing could simply be added as needed.    (015)

> So, I could start listing where all the technologies we've discussed fit 
> in, but I think that I best leave this post at this moment of clarity.  
> TouchGraph by the way might appear to only satisfy the second 
> requirement, but there is an easy way for it to handle sequence as 
> well.  My idea is to add special sequential edges (just a type of 
> directed edge).  These edges would function as trails on a map, "many 
> ways to get from point A to point B, but here is some advice from a 
> person who lives here."
> Ok, and with that plug I am out.    (016)

Well, as you know of course I'm using TouchGraph so far for graph
views. I think the only thing necessary to express timelines in
TG would be to have node widths match event length and pop them
along a horizontal line. Or something like that. It certainly
doesn't seem like an insurmountable problem.    (017)

Murray    (018)

[1] "A Gentle Introduction to Ted Nelson's ZigZag Structure"
Murray Altheim                  <http://kmi.open.ac.uk/people/murray/>
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK    (019)

      In the evening
      The rice leaves in the garden
      Rustle in the autumn wind
      That blows through my reed hut.  -- Minamoto no Tsunenobu    (020)