[x3d-public] [Semantics] FW: Dispense of top-level semantic relationships quickly!
yottzumm at gmail.com
Thu Jun 13 09:31:03 PDT 2019
Sent from Mail for Windows 10
From: John Carlson
Sent: Monday, June 10, 2019 4:17 AM
To: semantics at web3d.org
Subject: Dispense of top-level semantic relationships quickly!
Sorry, I joined late. I did post a few things to x3d-public. I think before we invent the future, we need to forget what we know, and look into the past (paraphrase from quote of Alan Kay):
1. We should use the UML class diagram relationship types (and potentially other diagram elements) as top level relationships between classes. We get IS_A and HAS_A from this. https://creately.com/blog/diagrams/class-diagram-relationships/ This will allow us to handle things like XML and X3DUOM schemas, I believe. This would also give us stereotypes from UML, or special icons for classes (I think the was a goal at one point). We do not need to reinvent domain specific visual languages—that is done with stereotypes in UML. Unless of course, you want to create a 3D-4D DSVL.
2. We should use Conceptual Dependency (from Shank, 1969) https://en.wikipedia.org/wiki/Conceptual_dependency_theory
https://deepblue.lib.umich.edu/bitstream/handle/2027.42/30278/0000679.pdf;jsessionid=5045F32362513C64F4FBFC0BA441CB6C?sequence=1 at least the set of primitive actions (predicates) which are intended for describing the real world in natural language. That is, natural language is close to semantics and how the layman describes reality. This will allow us to describe the semantics of animations which depend on primitive actions (animation “scripts”).
1 and 2 comprise the “top level” of any more concrete or domain-specific predicates. Can we list 10-20 predicates that all of our subsequent predicates will inherit from, and the relationships between the 10-20 predicates?
Are we discussing the semantics of dynamic and static scenes? Do we need separate semantics, or combined semantics?
I do not think that Shank had the answer in 1969. There’s a whole lot more if you consider eco-systems, networks, and worlds. (Actors comes to mind). There’s especially a desire to describe emergent behavior, like eating ½ an apple (thanks to Bart Kosko), breaking a leaf apart (a dream of mine), or dragging one’s hand through the water (thanks to Ted Nelson). How do you describe exceptions in terms of semantics?
Do we want to be able to query large databases of video, such as is used in developing Tesla’s autopilot software? What are people using for that?
I expect us to blow off the bottom of these stages in complexity, but first, let’s beg, borrow or steal a top level framework.
Do people feel we should work with a Model-View-Controller or Input-Processing-Storage-Output pattern first? Are we merely talking about View-Controller and Input-Output, or is there some Model/Processing/Storage involved? Shapes and Transforms come to mind for Models. Can we toss all our data into 5 bins? Can we reuse these names, or do we need names like Sensor, Node, Script, Route? Are there cross-bin classes?
I have a particular interest in creating a Taxonomy of Sensors or Controllers, if no one else is taking this. Does anyone have references beyond: https://www.nap.edu/read/4782/chapter/16 https://en.wikipedia.org/wiki/Robotic_sensors#Classification I think that adding sensor groups, group/individual oriented sensors and discrete/continual (Thanks Leonard Daly)/continuous sampling are important properties to express.
I can share my knowledge with the group RE: Hypermedia, Stocamedia, Chaomedia, and Quantumedia, and the 3 primitive media types, if needed. These are mainly focused on data generators. In my current role, I am a Senior Systems Generator (trying to go across systems paradigms). Mostly I generate scripts in the short term, but hope to extend this to languages, systems and networks as I work with Gina-Lauren or Christoph Valentin. In a previous life, I worked on Capacity Planning Systems.
As I generate additional systems, I expect the input devices to catch up to me, and then the ways that users and robots interact with the system.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the x3d-public