Embodied Language

[Linas Vepstas]

Attached is a pre-pre-draft of a description of what the current embodiment-language system does, including a walk-through of the current code, and how that code could be extended.

It might be that I am too early in sending this out, and I should have written more first. C'est la vie.  Release early, release often is the standard open-source mantra.

--linas

[Ed Pell]

Linas, documentation is even better than raw code for disseminating ideas/information in open source projects. Your document is much appreciated.

I am happy to see someone talking about models of the world and self.

I am also happy to see documentation that is less mathematics based and more readable.

Cheers,
Ed

[Ed Pell]

Linas, I would added a third model other. The three would be self, world, other. Where "other" my have two versions a generic other (person) and many copies of specific other (Alice, Bob, ...). Yes, other could be part of the world model but it feels special enough to me to have its own category. Specific other models can be based on profession, religious orientation, location, etc. That is we have internal models of "high level executive of large corporation", "school teacher", "southerner", "French person", etc.

Cheers,
Ed

[Linas Vepstas]

Yes, good point. I will add that.

--linas

[Noah Bliss]

For sanity's sake I greatly appreciate the indiscriminate application of the system to both self and world. I am not completely convinced this is how humans function as we definitely have certain delf-oriented instincts, but it will be interesting with this more "objecective" structure what kind of behaviors emerge. As for the "others" argument, I think it depends on how much we want to preteach it. If we truly wanted it to rely on structures gathered from pattern matching, then we should not discriminate between others and world, heck we don't even really need to distinguish between between self and world at least as far as design philosophy is concerned.

Sure at first with simple systems, its "self" is easy to distinguish, its body, its host machine, etc. But as we drive integration, the lines could easily blur, especially as it develops new methods of interacting with the world. E. g. if it matched patterns in markets and learned to create accounts to link into market APIs which it directly interacted with, then one could argue that that integration system it developed and uses is a part of its "self." If it were able to match market patterns well enough to get high accuracy and control, that could arguably be considered an extension of "self."

With a system built on the concepts opencogprime is, yes there is an incredible cleanliness and simplicity to it, but a lot of our human concepts don't apply cleanly.

[Ed Pell]

Noah, All, evolution gives humans and other living things many pre-canned functions like edge detectors in the eye. In vision system some people add the early edge detectors and some ask the blank slate to figure everything out for itself from scratch. My preference is to give the system as much of a leg up as possible. Later when we have functioning system we can explore the growth from blank slate question.

Self and not-self is a very early learning. I am that which I can control the movement of. The other is the warm, soft, and giving entity (Mom and Dad). The other becomes the family. Then strangers are discerned and apparently rated as dangerous instinctively, then learned as safe and not-safe strangers.

If we want to be experimental psychologists we can force the system to learn these early lesson from scratch. If we want a functioning embodied conversationalist in a reasonable amount of time then let's give it a head start. Both are valid pursuits. We have to pick one to start with.      

[Linas Vepstas]

For your reading pleasure, here is version 0.02 of the document. It details the language processing pipeline.  Please review, as it sets the stage for the next installment.

Please keep in mind that this document is describing the current code base, and will be raising design issues for the near-short-term design+implementation. As such, it is only philosophical or metaphysical to the point of how that affects specific design decisions.

Please keep in mind that the current code base *actually works*, and, on a good day, can be demoed.

If there are any questions, or any parts that are unclear, or vague, or obtuse, or seem to involve stupid design decisions, please raise these as issues.  I'll try to fix them.

--linas

--
You received this message because you are subscribed to the Google Groups "opencog" group.
To unsubscribe from this group and stop receiving emails from it, send an email to opencog+unsubscribe@googlegroups.com.
To post to this group, send email to ope...@googlegroups.com.
Visit this group at https://groups.google.com/group/opencog.
To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/7eacea5d-a872-4941-a4de-ba0af793e96e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[AmeBel]

Thanks Linas,

that is extremely helpful. Can't wait for 0.03 :-)

On Thursday, September 15, 2016 at 11:44:18 AM UTC+8, linas wrote:

For your reading pleasure, here is version 0.02 of the document. It details the language processing pipeline.  Please review, as it sets the stage for the next installment.

Please keep in mind that this document is describing the current code base, and will be raising design issues for the near-short-term design+implementation. As such, it is only philosophical or metaphysical to the point of how that affects specific design decisions.

Please keep in mind that the current code base *actually works*, and, on a good day, can be demoed.

If there are any questions, or any parts that are unclear, or vague, or obtuse, or seem to involve stupid design decisions, please raise these as issues.  I'll try to fix them.

--linas

On Wed, Sep 14, 2016 at 8:49 PM, Ed Pell <edp...@gmail.com> wrote:

Noah, All, evolution gives humans and other living things many pre-canned functions like edge detectors in the eye. In vision system some people add the early edge detectors and some ask the blank slate to figure everything out for itself from scratch. My preference is to give the system as much of a leg up as possible. Later when we have functioning system we can explore the growth from blank slate question.

Self and not-self is a very early learning. I am that which I can control the movement of. The other is the warm, soft, and giving entity (Mom and Dad). The other becomes the family. Then strangers are discerned and apparently rated as dangerous instinctively, then learned as safe and not-safe strangers.

If we want to be experimental psychologists we can force the system to learn these early lesson from scratch. If we want a functioning embodied conversationalist in a reasonable amount of time then let's give it a head start. Both are valid pursuits. We have to pick one to start with.      

--
You received this message because you are subscribed to the Google Groups "opencog" group.

To unsubscribe from this group and stop receiving emails from it, send an email to opencog+u...@googlegroups.com.

To post to this group, send email to ope...@googlegroups.com.
Visit this group at https://groups.google.com/group/opencog.

[Linas Vepstas]

On Thu, Sep 15, 2016 at 3:30 AM, AmeBel <am...@hansonrobotics.com> wrote:

Thanks Linas,

that is extremely helpful. Can't wait for 0.03 :-)

Here it is!  The centerpiece here is the diagram on page 11, with the backing text on pages 9 and 10 

Some confusing text on the (new) page 14 was fixed up.

It will be a few more days before I generate the next version. The stuff that was mostly easy to write was written, the next parts will take more effort.

--linas

[Ben Goertzel]

Thanks Linas!

I know putting together this sort of quality documentation takes a lot
of time and effort; but as you know it's critical for enabling others
to assist with extending your work on this stuff...

My current thinking is that in the next phase we can have concurrent
work on learning-based and hard-coding-based extensions of the current
prototype work. But let's discuss this in depth once you've completed
the document...

-- Ben

> --
> You received this message because you are subscribed to the Google Groups
> "opencog" group.

> To unsubscribe from this group and stop receiving emails from it, send an
> email to opencog+u...@googlegroups.com.
> To post to this group, send email to ope...@googlegroups.com.
> Visit this group at https://groups.google.com/group/opencog.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/opencog/CAHrUA34MQdz7u4Gsuj4%2BmpPJj%2BhLu-NVX0gOZ0FrkwxHemNgVg%40mail.gmail.com.

>
> For more options, visit https://groups.google.com/d/optout.

--
Ben Goertzel, PhD
http://goertzel.org

Super-benevolent super-intelligence is the thought the Global Brain is
currently struggling to form...