Virtual Humans Forum
Virtual Humans Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password
Forgot your Password?

Virtual Humans
Partner website:
Chatbots.org
 All Forums
 Virtual Humans
 Artificial Intelligence
 human mind map
 New Topic  Reply to Topic
 Printer Friendly
Previous Page | Next Page
Author Previous Topic Topic 
Page: of 25

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 20 2007 :  18:17:23  Show Profile  Reply with Quote
quote:
Originally posted by toborman


This type of statement tells harry how to identify the intent of the user.


Very nice Tom. My approach was a simple replacement of strings so that the input was simple restated in a standard form. Your approach could be more elegant (the learning aspect bound into the "means" concept) but I am not sure that it is functionally different than a simple replacement. It could well be very different if, somewhere else in the system, there is the option for a choice to make the change, or (more importantly) to evaluate the semantics (that word is not quite all of what I mean) of the conversation and adjust the meaning. Now that would be a clear step forward.

Uncle Jim (e=mc2)
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 20 2007 :  21:16:53  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
by jim
Very nice Tom. My approach was a simple replacement of strings so that the input was simple restated in a standard form. Your approach could be more elegant (the learning aspect bound into the "means" concept) but I am not sure that it is functionally different than a simple replacement. It could well be very different if, somewhere else in the system, there is the option for a choice to make the change, or (more importantly) to evaluate the semantics (that word is not quite all of what I mean) of the conversation and adjust the meaning. Now that would be a clear step forward.


Following the intention step in the conversation comes interpretation. Using this statement type harry can interpret clauses to his internal (mental) representation.

“*s are *s. “ in English interprets to “set.%2.contains all.set.%1.”
“* is a *. “ in English interprets to “set.%2.contains the.element.%1. ”
“*s are not *s. “ in English interprets to “set.%2.contains the.element.%1. ”
“*s are *. “ in English interprets to “attribute.%2.describes all.set.%1. ”
“*s are not *. “ in English interprets to “attribute.%2.describes not all.set.%1. ”
“some *s are not *s. “ in English interprets to “set.%2.contains not some.set.%1. ”
“some *s are *. “ in English interprets to “attribute.%2.describes some.set.%1. ”
“some *s are not *. “ in English interprets to “attribute.%2.describes not some.set.%1. ”
“some *s are *s. “ in English interprets to “set.%2.contains some.set.%1. ”
“*s can * *s. “ in English interprets to “set.%1.can %2 all.set.%3. ”
“*s can * *. “ in English interprets to “set.%1.can %2 all.attribute.%3. ”
“*s cannot * *s. “ in English interprets to “set.%1.can %2 not all.set.%3. ”
“*s cannot * *. “ in English interprets to “set.%1.can %2 not all.attribute.%3. ”
“* cannot * *. “ in English interprets to “element.%1.can %2 not all.attribute.%3. ”
“* cannot * *s. “ in English interprets to “element.%1.can %2 not all.set.%3. ”
“* can * *. “ in English interprets to “element.%1.can %2 all.attribute.%3. ”
“* can * *s. “ in English interprets to “element.%1.can %2 all.set.%3. ”
“* is not a *. “ in English interprets to “set.%2.contains not the.element.%1. ”
“* is not *. “ in English interprets to “attribute.%2.describes not the.element.%1. ”
* is *. “ in English interprets to “attribute.%2.describes the.element.%1. ”

http://mindmap.iwarp.com
Go to Top of Page

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 21 2007 :  00:18:41  Show Profile  Reply with Quote
quote:
Originally posted by toborman


Following the intention step in the conversation comes interpretation. Using this statement type harry can interpret clauses to his internal (mental) representation.



Very nice. Very.

I use fuzzy logic to shade the interpretation of the final input and identify response options given a personality filter for the VH. Multi-valued set theory (fuzzy logic of course) can be extremely useful in interpreting input and shading output. But frankly I don't know that you need it from what I see.



Uncle Jim (e=mc2)
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 21 2007 :  02:10:16  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
by jim
I use fuzzy logic to shade the interpretation of the final input and identify response options given a personality filter for the VH. Multi-valued set theory (fuzzy logic of course) can be extremely useful in interpreting input and shading output. But frankly I don't know that you need it from what I see.



I haven't found a place for fuzzy sets yet, but I'm sure it's just around the corner.

http://mindmap.iwarp.com
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 21 2007 :  05:05:53  Show Profile  Visit toborman's Homepage  Reply with Quote
After intent and “meaning” have been established the “director” passes control to the appropriate “role” or function. The information is processed and responses are generated. If the “role” wants to control the conversation the director is notified. Responses are consolidated, translated to natural language (English), transformed to correct grammar and syntax, and spoken by the conversationalist.

Harry is taught to handle these functions with the language file. For example, consolidation is controlled with this statement type.

“*.*.describes the.*.*. and *.*.describes the.*.%4.” consolidates to “attribute.%2_ %6.describes the.element.%4.”
“*.*.describes not the.*.*. and *.*.describes not the.*.%4.” consolidates to “attribute.%2_ %6.describes not the.element.%4.”
“*.*.contains the.*.*. and *.%2.contains the.*.*.” consolidates to “set.%2.contains the.elements.%7_%4.”
“*.*.contains not the.*.*. and *.%2.contains not the.*.*.” consolidates to “set.%2.contains not the.elements.%4_%7.”
“*.*.contains the.*.*. and *.*.describes not the.*.%4.” consolidates to “set.%6_ %2.contains not the.element.%4.”

http://mindmap.iwarp.com
Go to Top of Page

GrantNZ
Dedicated Member



New Zealand
2677 Posts

Posted - Dec 21 2007 :  08:18:59  Show Profile  Reply with Quote
I like the design.

Just out of interest, in what language are you programming?
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 21 2007 :  12:20:40  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
by grant
Just out of interest, in what language are you programming?


I'm using VBA with MS Access.

http://mindmap.iwarp.com
Go to Top of Page

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 22 2007 :  18:38:59  Show Profile  Reply with Quote
quote:
Originally posted by toborman

quote:
by grant
Just out of interest, in what language are you programming?


I'm using VBA with MS Access.


Well, my guess was wrong, darn. Glad I didn't bet anybody on it. I was guessing you were using RealBasic because I saw their logo on one of your data files in the html download.

Uncle Jim (e=mc2)
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 22 2007 :  20:38:15  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
The only thing I can suggest, I guess, is that you try to devise your AI around a dynamic architecture as much as possible. The rules of the English language can be very cryptic so you want to stay away from hard-coding as much as possible... that's what I've learned anyway. When someone talks to someone else in English... well its crazy keeping track of all the rules implied and that we take for granted... so you don't want to lock in your AI to work only if someone words their sentences in a certain way.



You are right, of course. All natural language changes over time, and across social groups. Hopefully I will get to the point in the process that will accomodate those changes through inductive reasoning. In the meantime, Harry's files can be changed through the tutoring process.

Harry uses this statement type to translate his mental representation to English.

Translate ”attribute.*.describes the.element.*.” into English ”%2 is %1. ”
Translate ”set.*.contains the.element.*. ” into English ”%2 is a %1. ”
Translate ”set.*.can * all.set.*. ” into English ”%1s %2 %3s. ”
Translate ”set.*.can * all.attribute.*. ” into English ”%1s %2 %3. ”
Translate ”element.*.can * all.attribute.*. ” into English ”%1 %2s %3. ”
Translate ”attribute.*.describes not all.set.*. ” into English ”%2s are not %1. ”
Translate ”attribute.*.describes not the.element.*. ” into English ”%2 is not %1. ”
Translate ”set.*.contains not the.element.*. ” into English ”%2 is not a %1. ”
Translate ”set.*.can * not all.set.*. ” into English ”%1s cannot %2 %3s. ”
Translate ”set.*.can * not all.attribute.*. ” into English ”%1s cannot %2 %3. ”
Translate ”element.*.can * not all.attribute.*. ” into English ”%1 cannot %2 %3. ”
Translate ”element.*.can * not all.set.*. ” into English ”%1 cannot %2 %3s. ”

http://mindmap.iwarp.com
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 23 2007 :  01:49:39  Show Profile  Visit toborman's Homepage  Reply with Quote
It’s nice that Harry can handle natural language as a conversationalist, but his real intelligence is better represented in his other roles. First, Harry needs to remember and recall facts provided by the user. Then Harry needs to be able to figure out new facts things from the facts it knows. This process of figuring out new things might be triggered when Harry is asked a question that he can’t answer with facts he knows. To accomplish this he must have rules that allow him to infer new facts. These are the rules of deduction, induction, and abduction. Harry’s “learner” role allows him to remember, recall and infer new facts. The “learner” is also able to say how he learned the facts.

Harry : how are you.
Harry : it is 03:54 PM, Saturday, December 22, 2007.
tom : jim is a human.
Harry : I will remember.
Harry : jim is a human.
tom : is jim an animal?
Harry : I don't know if the statement is true.
tom : humans are animals.
Harry : I will remember.
Harry : humans are animals.
tom : is jim an animal?
Harry : Yes, jim is an animal.
tom : prove jim is an animal.
Harry : because humans are animals and jim is a human, it follows that jim is an animal.
tom : (OK)
Harry : it's been a pleasure.

http://mindmap.iwarp.com
Go to Top of Page

GrantNZ
Dedicated Member



New Zealand
2677 Posts

Posted - Dec 23 2007 :  02:58:42  Show Profile  Reply with Quote
I say again, I like the design. Having an AI explain his deductions is important to me too.

quote:
This process of figuring out new things might be triggered when Harry is asked a question that he can’t answer with facts he knows.

I like the idea of "dream time" - i.e. the AI using idle time to analyse all the data to infer or ab/deduct anything it can.

This could be done on the spot too. So if you add the remark "animals are alive," Harry could answer with "that means humans are alive." or "does that mean jim is alive?"
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 23 2007 :  03:50:39  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
by grant

I like the idea of "dream time" - i.e. the AI using idle time to analyse all the data to infer or ab/deduct anything it can.


Great idea. Then Harry could relate his new discoveries later.

quote:
by grant

This could be done on the spot too. So if you add the remark "animals are alive," Harry could answer with "that means humans are alive." or "does that mean jim is alive?"


I'll have to teach Harry to be more inquisitive.

http://mindmap.iwarp.com
Go to Top of Page

toborman
Hooked Member



USA
289 Posts

Posted - Dec 23 2007 :  18:40:27  Show Profile  Visit toborman's Homepage  Reply with Quote
Now we need to teach Harry to find a previous statement to answer the question “why”.

tom : hello. what time is it?
Harry : how are you.
Harry : it is 09:30 AM, Sunday, December 23, 2007.
tom : is jim an animal?
Harry : Yes, jim is an animal.
tom : why?
Harry : since humans are animals and jim is a human, it follows that jim is an animal.
tom : (OK)
Harry : it's been a pleasure.

http://mindmap.iwarp.com
Go to Top of Page

kenkirkland
Curious Member



58 Posts

Posted - Dec 23 2007 :  19:25:02  Show Profile  Reply with Quote
Great Work!!

Does Harry handle "prove" differently than "why"?

When Harry is asked "is jim an animal?", is Harry remembering his prior answer or is he rethinking the answer?

Ken
Go to Top of Page

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 23 2007 :  22:59:25  Show Profile
quote:
Originally posted by GrantNZ

So if you add the remark "animals are alive," Harry could answer with "that means humans are alive." or "does that mean jim is alive?"


I am waiting with roasted-garlic-chicken pizza "baited" breath for his answer.

Uncle Jim (e=mc2)

Edited by - laackejim on Dec 23 2007 23:00:15
Go to Top of Page
Page: of 25 Previous Topic Topic   
Previous Page | Next Page
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
Virtual Humans Forum © V.R.Consulting Go To Top Of Page
This page was generated in 0.14 seconds. Snitz Forums 2000