[x3d-public] Ways to generate X3D for testing?

John Carlson yottzumm at gmail.com
Sun Feb 18 23:22:58 PST 2018

What is the potential for generating XML or VRML or JSON with neural networks? I assume hierarchies/graphs could be done with something like:



Sent from Mail for Windows 10

From: John Carlson
Sent: Sunday, February 18, 2018 11:35 PM
To: x3d-public at web3d.org
Subject: Ways to generate X3D for testing?

What ways have people used to automatically generate X3D (VRML) for testing?

I can think of a few approaches:

1. Schema or language driven fuzzy testing, possibly with a probabilistic language or schema.
2. Assume the X3D is being generated by separate sensors, one per tag or class.  Sensors or queue servers will also generate references between inputs to create a hierarchy or graph. References will have a type associated with them. Fragments will be bound at references with DEF or USE attributes, if possible. Program the sensors and queue servers with existing X3D files. Thus X3D files will be emulated/simulated with multisensor input data with references between the input data.   This might be similar to a discrete event model generating work for a system.  Brian, is this the idea behind the universal translator?
3. Same as 2, but everything is thrown into a soup instead of queues, and the servers are given locomotion and grasping capabilities.  I call this the “Enzyme Approach.”  I believe this may be a P-system approach as well.
4. Some kind of combination between 2 and 3. Unsure what this is so far.
5. Do some kind of 3D image reconstruction/vision software.
6. Others?

I am starting to take the approach 2, and wonder if anyone has feedback on the approach, or any other approach used, especially for generating good test cases.  References on automatically generating XML examples are welcome.

One thing I am think of is that I will need multiple queue servers for each number of children of a particular node, all serving the same queue.   What do you think?

Any output from the above techniques will be checked with schema, schematron and validated with X3DJSAIL.  It is possible that the result of validation will be fed back into the X3D generation programs (how?).

If I do do 2, how can I possibly tune the queues for best performance?



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20180219/de9d9cf1/attachment-0001.html>

More information about the x3d-public mailing list