RE: [agi] Early Apps.

From: Gary Miller (garymiller@starband.net)
Date: Sun Dec 29 2002 - 17:22:34 MST


Hi Jonathon,
 
You said...
 
>> that's a really good idea, I bet it has proved very useful...
 
>> do you have a website with a demo of your project?
 
I hope you do not mind but I am copying my reply to you to the mailing
list since I have had similar questions from others on the list and am
giving you a few examples that some of the others may find interesting
and applicable to their AGI research.
 
Sorry Jonathon, I do not yet have a website for my project. I am afraid
my hosting costs would be very high. Right now I require at least a
2.2 Ghz machine with 512MB of memory running .Net runtime to give
satisfactory run time for a single user! I also require about 300MB of
disk to hold the Knowledge Base, executables and leave room for some log
files. The only operating system I have tested this on yet is Windows
XP Professional. Each web client would need to run a separate instance
of the application also due to the bot's need to maintain context within
each conversation.
 
My user interface is now a console mode screen to meet the Loebner
competition requirements but since it's written in VB.Net I don't think
it will be difficult to turn it into a web service. I have some college
students giving me a bid on a Web interface and trying to get me and
inexpensive host site, but the last estimate I received from a national
ISP was about $750 a month.
 
I had hoped to get something up by March for the Chatterbot challenge
contest, but that will depend on what kind of cost and time my estimate
comes back with. The number one priority for my time still has to be
increasing the base intelligence of the system.
 
Feel free to ask any questions you may have though and even if I can't
open my website up to everyone when it's done due to processing
constraints, I'll be sure to contact you and allocate some time for you
on the system.
 
I like your website. I see you are into immortality also. I read and
was very intrigued by the first part of your immortality how-to! Will
you include your views on dietary restriction, antioxidant and nutrient
supplementation as well? Please drop me a line when you complete it.
You should also consider publishing it on www.imminst.org they are a
site that is devoted to immortality.
 
You may already be aware of this site, but just in case you aren't I
thought I would mention it to you also.
 
http://www.find-a-drug.com/
 
You must have a very interesting background you seem to have a lot of
background in neural systems. I read through some of your other AI
articles and got lost very quickly. I would be interested to see your
ISL spec when it's completed though. The closest thing I have to your
ISL are my embedded pattern variables which contain information
regarding set membership:
 
In English: Cats, dogs, and rats are mammals
 
In NPL: [Mammal]="{cat|dog|rat}"
 
If input from user: "A cow is a mammal"
 
Then NPL becomes [Mammal]="{cat|cow|dog|rat}"
 
Where [Mammal] is the pattern variable name.
          {} says one of the enclosed set members is required
          | indicates a "logical or" set member seperator
 
So if the user asks "Is dog a mammal"
 
Pattern 1 << Is a [Mammal] a mam(m){a|e}l[eoln]
Response 1 >> Yes a [#Token3] is a mammal
 
Pattern 2 << Is a * a mam(m){a|e}l[eoln]
Response 2 >> No, #proc("What is a "[#star1])
 
Patterns priority is controlled and pattern 1 will attempt to match
before pattern 2 so...
 
For any known mammal pattern 1 will be satisfied and the response Yes, a
dog is a mammal will be emitted.
 
For any other string such as "Is a chair a mammal?" the bot will say
"No," and then ask itself the question
"What is a chair" which would recurse another level down through the
pattern matcher with the new question and
append it's answer onto the "No," from the first level match so that the
final response might be "No, a chair is a piece
of furniture used to sit on."
 
Note the [#Token3]], [#star1], and #proc("What is a "[#star1]) are
functions used to reformulate thoughts based on information contained in
the original user input.
 
Note the mam(m){a|e}l in the pattern allows mammal to be spelled mamel,
mamal, mammal, or mammel to allow for common misspellings.
An additional Levenschtein distance calculation is also used if a less
common misspelling is encountered and the pattern does not match pattern
1.
 
This is an extremely simple example. My complete pattern language
definition can get complex allowing all the ways of expressing a single
thought to be encapsulated into a single pattern. The average pattern
length allowing for synonyms and variations in sentence structure runs
about 320 characters. But some such as the pattern for greetings can
run 2400 characters in length to accomadate all of the various ways of
greeting a person.
 
I hope this gives you a taste of my research and would enjoy hearing
more about your research and background. I did not find any
biographical info on yourself on your website
 
Best Regards,
    Gary A. Miller

-----Original Message-----
From: Jonathan Standley [mailto:standley@rcn.com]
Sent: Sunday, December 29, 2002 1:44 PM
To: garymiller@starband.net
Subject: Re: [agi] Early Apps.

 
>
> I view what I am doing distinct from expert systems because I do not
yet
> use either a backward or forward inference engine to satisfy a limited
> number of goal states. The knowledge base is not in the form of rules
> but rather many matched patterns and encoded factoids of knowledge
many
> of which are transitory in nature and track the context of the
> conversation. Each pattern may trigger a request for additional
> information like an expert system. But the bot does not have a
> particular goal state in mind other that learning new information
unless
> a specific request of it is made by the user. I also differ from Cyc
in
> that realizing the importance of English as a user interface from the
> beginning, all internal thoughts and goal states occur as an internal
> dialog in English. This eliminates the requirement to translate an
> internal knowledge representation to an external natural language
other
> than providing one or response patterns to specific input patterns.
It
> also makes it easy to monitor what the bot is learning and whether it
is
> making proper inferences because it's internal thought process is
> displayed in English while in debug mode..
 
that's a really good idea, I bet it has proved very useful...
 
do you have a website with a demo of your project?
 
J. Standley
http://users.rcn.com/standley/AI/AI.htm
 
 



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:41 MDT