Copyright 2008 David Dodds
Conceptual Metaphor and the Ontoclock System

My first paper on conceptual metaphor processing was for the January
1981 AAAS meeting in Toronto, it also had alot of Zadeh / Goguen Fuzzy
Systems components.

There is a paper which I got recently from Univ Orgeon lib which shows
the math behind a hierarchical homuncular system, the authors called
it BACON-H, which was a modification of the BACON system. For many
domains this is a pretty good system, it learns and is able to form
some metaprogramming capabilities.

There are various kinds of so-called discovery systems, BACON is just
one. Its focus is on 'discovering' mathematical equations (they call
them 'laws' gufaw) which map the data sets given to the program as
input into a regularity (an equation which is used as a description of
the regularity).
There is no discovery of course, BACON is simply algorithmic, and its
input data cannot be too noisy or BACON cant tell its butt from a ...
Also BACON has to have a smallish number of variables, it isnt able to
winnow or weed out extraneous variables worth a hoot. This is the most
egregious thing I saw about BACON when I first came across it years
ago. Momentarily pleased about the program's capabilities, having just
read the author's glowing writeup about it, I fell to feeling disdain
for it once I saw how the program worked and what it really was. (a
crock, perpetrated by a renowned researcher).
In the years since, I have looked at studies of 'data mining'
technology and seen much more sophisticated 'discovery' programs /
processing.

In 2007 I presented a paper, 'The Difference Between Having Knowledge
and Knowing It', which used an SVG program called Ontoclock to
demonstrate the main ideas of the presentation.
In the case of the Ontoclock system it just so happened that
understanding' involved was do-able through some logic and
mathematical means. The data was extremely clean, because it was
generated by a strict algorithm not data from sensors in the real-world.
The ontoclock system was not set up to look at absolutely all the code
(in it), nor to understand same and hence it doesnt 'understand' that
the variables that it IS monitoring constitute pieces/parts of a
larger whole (what I have called here 'a strict algorithm'(ic process)).


In case you dont have the ontoclock code handy here is what I was
referring to by "generated by a strict algorithm not data from sensors
in the real-world".
The code for that algorithm is next:

function getTime ()
{
var Now = new Date();
lastSeconds = Seconds;
lastMinutes = Minutes;
lastHours = Hours;
Seconds = Now.getSeconds();
Minutes = Now.getMinutes();
Hours = Now.getHours() + hroffset + Minutes / 60;
}


function showTime ()
{
getTime ();


document.getElementById("seconds").setAttribute('transform', 'rotate('
+ (Seconds * 6) + ')');

if (Minutes != lastMinutes) {

document.getElementById("minutes").setAttribute('transform', 'rotate('
+ (Minutes * 6) + ')');
}
if (Hours != lastHours) {

document.getElementById("hours").setAttribute('transform', 'rotate(' +
(Hours * 30) + ')');
}
}



When we are talking about 'seconds' (of time) and the 'sweep second
hand of the clock' (picture) the key parts, seems to me, are:

var Now = new Date();
Seconds = Now.getSeconds();
document.getElementById("seconds").setAttribute('transform', 'rotate('
+ (Seconds * 6) + ')');



the discovery monitor is 'watching' the variables/functions
Date()
Now
Now.getSeconds()
Seconds
Seconds {the same Seconds as just prior,
but located in 'document.getElementById("seconds")' statement}


what is 'discovered' is that the SVG object called
document.getElementById("seconds")
is defined as having the (new) position angle=zero + (Seconds * 6), for
each (successive or 'new') Seconds value.
'setInterval ("showTime ()",1000);' is the thing that amounts
to 'the clock tick which occurs once per second'
One may wish to view the setInterval statement as a kind of interrupt.
(If one really has to go further back than that then one could
consider a second to
be such-and-such-many counts of cycles of a cesium clock beam.
All the seconds are pretty much the same and there is no particular
time corresponding
to a particular second, unless one counts the sequence of 'second'
starting at zero.)

It is also discovered that the new position is always 6 degrees later'
(or clockwise) than the previous position was. Since the 'change' is
'rotation' this implies a circular 'motion' (change in position (with
some dependent variable (time in this case))).
Knowledge in the system (via ontologies) defines a circle as 'having'
'angles' and that there are 360 equally spaced such angles
constituting a complete circle. A bit of (dodgy) BACON-type processing
here discovers that the ontology defines:
1 minute = 60 seconds . The count for Seconds never goes larger than
60 and therefore the angle calculation (of
document.getElementById("seconds").setAttribute('transform', 'rotate('
+ (Seconds * 6) + ')');
never exceeds 360 and that 360 is the largest possible 'legal' or
'correct' value for an angle of a circle. It is discovered that the
'sweep second hand' 'jumps' (ie 'changes location') clockwise by 6
degrees once per second. The value 6 is constant, it never changes.
It is inferred that since there are 360 degrees in a circle and a
range of seconds from 0 to 59 (?) [a count of 60] {and it is magically
implied from somewhere that the movements of the 'clock hands' are to
be via equally spaced angles of rotation for each change of
hand-position} then each change of hand-position should be 360/60
degrees.
(Thats 6 degrees, hence the multiplication by 6 is 'explained' by this
discovery.) (Since there are the same number of minutes in an hour and
the minutes portrayed in an hour pass through 360 degree circle they
too are multiplied by 6 and the other considerations are the same
because of that. There are 12 hours (in this clock, its not a 24 hour
clock) and therefore each hour 'has' a 360/12 = 30 degree 'rotation'
of the [hour-]'hand', 'on the (SVG) clock'.

Applying some 'logic' (and math) to the code above the computer can
'discover' that the 'hands' only 'move' ('change location') when the
numeric value of the rotation calculation (ie Seconds, Minutes, Hours)
changes. Also it is discovered that the Seconds hand, for instance,
never moves 1, or 2, or 3 or 4 or 5 degrees, only exactly 6.
This is a minor revelation to a computer, while the 'time occurs'
contiguously, from 0 through 59 seconds by 1's; the 'location' occurs'
only in steps of (exactly) 6!
With no 1, 2, 3, 4, 5 in-between! (This is the sort of thing that then
gets a rational being like a human thinking about: why are there 360
degrees in a circle, why are there 60 seconds in a minute, why are
there 60 minutes in an hour, why are there 12 (or 24) hours in a
half-day-AM-PM / day. The Aztecs used 12 for a lot of things, but
Captain Picard uses 10 for his stardate system.

The 'understanding' described in this episode is not as rich as it is
in humans, this is mostly because of two reasons: 1) there is no
subjective agent to appreciate (cognitively experience) the contents
of the 'understanding' and 2) the components of the 'understanding'
are rather sparse compared with those of humans and lacking in larger
numbers of referents, such as (associated) meta data and processes.