[semantics-public] interpreter -> compiler -> Humanoid Animation

[John Carlson]

Don, all, would it be possible to get the X3DUOM as RDF/Turtle (just as an
interesting exercise)?  Thanks!

Michael,

TL;DR;  Just read last paragraph if you like.

The subject basically describes all the technology I've been describing to
you.  I have never achieved a complete compiler to machine code for a
general language. It's possible that I've created a bytecode interpreter
using the translator flat file format converted to
compileable/interpretable C++ variable declarations. I did at one point,
expand branches into the steps a branch would take as a C++ function, but
went no further and backed out, primarily because I wanted to decompile the
code.  I was able to compare the flat file to the decompiled code by
re-persisting the programs, to verify the source code generator.

This is how I am a languages guy, I guess.  I am not a big Natural Semantic
Metalanguage (NSM) fan, but if NSM statements can be couched in the terms
of persistent objects or class grammar (even list of lists...of primes--the
obvious choice), and can fulfill the last paragraph, I would be interested
in exploring the NSM concept for whatever purpose you want.

As a side note, I state that I've been able to convert documents to lists
of lists of words.  I merely used something like PDF->HTML, and converted
HTML div's to JSON (JavaScript Object Notation) arrays. I wanted to do JSON
translation-by-demonstration and I had an existing data source.  We can
have a very large supply of documents represented as JSON arrays, if we
need it. I realize that NSM-DALIA takes English.

I guess we're dealing with a *semantic* object model and *semantic* graph
with NSM.  The Web3D consortium is currently working with RDF/Turtle I
believe.  There's also OWL/OWL2.  I should be able to provide you with
around 3000 .ttl (RDF/Turtle) files in a single domain (X3D) translated
from XML:
https://www.web3d.org/x3d/content/examples/X3dResources.html#Examples (try
Online link, .ttl is on right for individual scenes).  Did you send me an
NSM Bible at one point?  Can we translate RDF/Turtle to NSM?

Ah yes!
https://openresearch-repository.anu.edu.au/bitstream/1885/155252/2/What%20Christians%20Believe%20for%20Open%20Research%2020190204.pdf

I'm not saying that RDF/Turtle is even desirable.  I just have a bunch of
XML files (X3D scenes--non-RDF) I'd like to convert to their processable
semantics.  We may have to improvise some of the semantics, that is,
someone converts animations to semantics, such as "walk."  (I know this
isn't an NSM prime.  I want something like "move from point to point
upright at normal speed using legs") It's very likely we would have to have
thousands of walk examples, and many more with "not walk", and do some kind
of supervised learning with current technology that I've seen (I need to
think about adversarial networks, here).  It may be possible to convert
each NSM prime to an animation, IDK.  I tried both the dictionary approach
(word->video) and the SignWriting approach (word->icons)

I know NSM is about breaking down larger structures into simpler ones.  Say
I'd like to reduce the X3D to VR/AR animations (assume for conversion to a
4D printer/animatronics).  If NSM-DALIA would help with that, I'm all
ears.  If we can even create animations from NSM phrases (skipping the
X3D), that would be awesome!  Can we create an animation of Towers of Hanoi
from the NSM code in your Natural Semantic Programming paper?

Yes, I realize you assigned me that very goal with translating a single
word to an animation, which I have not achieved yet.

In other words, we need to extend NSM to handle virtual worlds, or
satellite worlds, not only human worlds, right? How does one describe a
virtual world in NSM?  We have Towers of Hanoi.   We can obviously use any
language we like as long as it can be reduced to NSM primes.

The HAnim (Humanoid Animation) ISO standard:
https://www.web3d.org/documents/specifications/19774/V2.0/index.html  has
been ratified.  We now need examples of HAnimMotion (.bvh) elements.  This
is my other job.

The difficulty in all of this is translating from spatiotemporal semantics
(geometry, coordinates, etc.) to/from NSM semantics. The NSM primes for
this are:

Time:  WHEN/TIME, NOW, BEFORE, AFTER, A LONG TIME, A SHORT TIME, FOR SOME
TIME, MOMENT

Space: WHERE/PLACE, HERE, ABOVE, BELOW, FAR, NEAR, SIDE, INSIDE, TOUCH
(CONTACT)

Thus, there are three kinds of "output," movies, symbols, and shapes. NSM
handles symbols.   What handles movies and shapes?  X3D!  I don't care if
it's VRML, XML, JSON, Turtle, Python, JavaScript, Java ... X3D is the NSM
of graphics!  Now, how can NSM and X3D work together?

John
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/semantics-public_web3d.org/attachments/20200716/2ccf6509/attachment.html>