Computing Sense of Spatiality 5

Copyright 2008 David Dodds


Remember we are looking at the concept "near". We previously looked at
the calculation which computes the direct linear distance
("separation") between two points. If we use the two text strings x,y
origins we can compute a crisp value of separation (of them). In the
bar chart illustration these two bar origins are 40 units apart along
the x-axis but since they are of different heights their actual
distance of separation depends on the heights of the bars and so bars
that might be otherwise close together or quite "near" (each other)
will be measured as farther apart "not so near" when one bar is "tall"
and the other is "short", if we are using the rectangle's x,y origin
as "location" (of the rectangle).

Look at the bar chart illustration. Bar 3 is tall and Bar5 is rather
short. There is only one 20 unit wide bar separating them and so in
that respect they are relatively close together even though the
origin-to-origin distance measure between the two bars numerically
suggests that they are not at all that close together. The problem is
in the way the "near" value was computed. It did not take into account
either the morphology of the two (bar) objects nor their orientation
with respect to the canvas origin axes.

What we need to do is include both of those sets of information in the
computation for "near".

In a next episode we will see the programming for how to do these.
Also next episode we will see discussion of the use of
meta-programming as a means for implementing concepts like "near"
("/far"), "above" "below", "beside", "in front", "in back" ("behind").
These concepts are so oft used by us as to seem obvious and our
familiarity through frequent use makes them seem near trivial. This
is because these concepts are processed in our subconscious and hence
any complexities involved in the processing of these concepts is not
available to our consciousness or awareness. We are consciously able
to deliberate on these concepts but in their typical day to day use we
only become conscious or aware of the results of this processing but
are not privy to the details or innards of the processing. In a sense
our consciousness could be likened to a programmer who submits or runs
parameterized functions to a processor. The innards and doings of the
processor are opaque, few users have knowledge of the logic circuitry
used by the computer to do addition or other aritmetic, but we don't
care as long as it seems to be working ok. I've just described the
average windows user mouse clicks and drag-and-drop.

The user in this scenario runs functions and programs on his PC (the
subconscious). He may not know about the innards of the code which is
associated with what he clicked on or dragged but he doesn't need to
know, as long as it effects the results he desires by activating it.
The windows user has a mental map which maps things he wants done with
things to click on and drag and drop. As long as his clicks and drags
accomplished what he wanted done he didn't care what was actually
being done or how. This is the (human) metaprogrammer.

In a future episode we will see a model of the human metaprogrammer
and how "knowing how" (to determine how-) 'near' Bar3 and Bar5 are to
each other is achieved programmatically. We don't have a TV camera we
can point at the barchart illustration nor do we have a machine that
can take the video data stream from such a camera and "see" the
illustration. Seeing (and understanding what is seen) is not just
photography.

In order that the computer system be able to "perceive" near-ness and
the other mentioned spatial relationships in a way which seems similar
or familiar to how we do we need to have the computer do its
processing such that the output or result is like our own perceptions.
It is absolutely not the case that the way the computer processing is
performed is the same as in us.

When we look at the bar chart picture we have a) a mental sense of the
picture, b) a linguistically mediated sense of the picture (ie
'words') and c) possibly some emotions are triggered (or possibly
not). The "a)" sense is the same kind of (mental) stuff that occurs
when we look out at the people in front of us as we walk along the
sidewalk. Humans are capable of applying linguistic processing to this
"a)" mental content and again the details or innards of that
linguification or turning non-linguistic mental content (thought) into
linguistic content are performed in the subconscious and we are not at
all aware of the details of the linguification processing itself we
are aware only of the (linguistic) [ie 'words'] output from it.

In a next episode we will see discussion of programming which
processes spatial information in a way that results in analyses
similar to how we perceive "near-ness", "above" etc. Part of the key
to such processing is to determine and use "context" in the
interpretation ("perception") of the spatial relations.
Metaprogramming / planning is discussed as a means to organize /
orchestrate the programs which perform the spatial analysis and
linguification of the results.

Next time also we read about Daniel Weld's book about how one decides
what to say (when talking or writing); also about metaprogramming,
reflection in programing and a bit about AspectJ an Aspect Oriented
Programming language.