[x3d-public] Parameterized VRML, X3D for i18n, l10n. How can we semi-automate create producing stylesheets or serializers for X3DUOM and X3D XML?

[John Carlson]

So I can see how one can take the X3DUOM (X3D unified object model) and 
create stylesheets for each target output language/library, whether 
that's X3DJSAIL, x3d.py, X3D JSON schema, field types, and parent/child 
method overrides (mapToMethod.js).

And I see how one can take X3D XML and produce Java, Python, JSON, 
JavaScript, X_ITE, X3DOM, etc. using stylesheets.

I peeked at the x3d/tooltips folder on sourceforge a bit. Looks like 
lots of translation work needs to be done.

How for example, how would I create a French X3DJSAIL or X3D-Edit?

Or how would I translate my X3D XML files to different natural languages?

One of the things I'm thinking of is providing different Shapes/Textures 
based on a natural language/semantics/culture. That is, will we begin 
translating semantics (.ttl) to natural languages/culture via DFDL or XSL?

For example, the textures or shapes might be different if you're in a 
Russian city versus an American city, but the underlying semantics, a 
city, remains the same. I've heard of the Chinese duplicating cities 
from other countries.

How does "style transfer" from ML/Deep Learning/AI affect what avenues 
we might take?  How do we "style transfer" XML?  Can we only do textures?

*How can we semi-automate producing stylesheets or serializers for 
X3DUOM and X3D XML?*

Or perhaps another approach might be possible we haven't thought of yet?

My feeble attempts at this include looking at XSLbyDemo by IBM, VXT, 
Jonathan Sprinkle's work and DFDL for short periods.  There's also 
technology for doing model transformation with XML.  I do not know if 
these are viable solutions or not, or if Altova provides a superior 
product.  I'm guessing that Altova is pretty much the top dog in this 
space still?

How do we capture the differences between Java, Python, JSON Schema 
object model translations (SAI bindings) and the differences between 
Java, Python, JSON, JavaScript, X_ITE, X3DOM object translations 
(encoding translations from XML?).

Can we combine object models with objects?  JSON schema is JSON, for 
example.  It might be possible to override schema in an document.

https://www.altova.com/xmlspy-xml-editor/xslt-editor looks like it 
contains "XSLT Back-mapping."  Interesting.

Is templating the answer?

Can we create a meta-model (xsd) from an object model (xml), and use the 
meta-model to direct object model translation?  How can we use XML 
Schema to direct XML Document translation?  Is this what DFDL is used for?

In general, it would be interesting to translate X3DUOM back to XML Schema.

Do we just need to do enough XSLT stylesheets to get an "aha" about how 
to proceed?

Well, I think I'll see if X3D-Edit will compile.

John

On 6/16/21 4:42 PM, John Carlson wrote:
> Has anyone done parameterized VRML, X3D or video for i18n or l10n?
>
> Thanks!
>
> John