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  Topic Next Topic
Page: of 25

toborman
Hooked Member



USA
291 Posts

Posted - May 01 2009 :  08:46:15  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
Hi guys, you're doing some great work And welcome aboard, Hatcheda!

Have you considered using a dictionary system (such as WordNet) for handling things like pluralisation? That may be an easy way to handle some language interpretation difficulties.

I think some of those dictionary things have set descriptions too, e.g. "cat" is a "mammal", which may give your AIs a head start. You may want to consider those descriptions low-quality, since I'm not sure if they're well arranged for AI use, but at least Harry or Cortona could answer "I think so" if you ask "is a cat a mammal," even if he has never talked about cats or mammals before!

Yes, I have checked wordnet. Harry may want to look things up at some point and when he does this is a good source of machine readable information.

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

Hatcheda
Curious Member



27 Posts

Posted - May 01 2009 :  09:27:08  Show Profile  Reply with Quote
Grammar.au3 addition (tonights work) wip
http://cid-dad9dc46d450d472.skydrive.live.com/browse.aspx/Cortona

Edited by - Hatcheda on May 06 2009 21:12:42
Go to Top of Page

toborman
Hooked Member



USA
291 Posts

Posted - May 01 2009 :  15:45:59  Show Profile  Visit toborman's Homepage  Reply with Quote
Harry is envious of Cortona's new abilities and reminds her that she'll have to translate singular to plural for some sentence structures on output.

With all this new code I'm wondering when you have time to sleep. If you're like me you're still thinking about this when you're not coding.

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

Hatcheda
Curious Member



27 Posts

Posted - May 01 2009 :  22:23:16  Show Profile  Reply with Quote
quote:
Harry is envious of Cortona's new abilities and reminds her that she'll have to translate singular to plural for some sentence structures on output.


It's here for you to translate if needed. (the code) With vb6 I think you would just need to dim your strings and such then adapt the functions used from their personal UDF's. They may be of much use elsewhere as well.
As for Singular to Plural. I can just reverse it, I think.

quote:
With all this new code I'm wondering when you have time to sleep. If you're like me you're still thinking about this when you're not coding.


quote:
David: Cortona, What is sleep?
Cortona: David, Sleep is the time you sacrifice to create Cortona.


no, not really :-) but soon enough, I hope.

I have a long drive to work, gives me plenty of time to think about my next step and how to approach it.

I am building Cortona to be as logical as possible. I do not mind creating the translations for her. English is not logical, and is not "Correct". As such, I will create the rules for her to follow, since they are not practical to approach through logic or reason. Perhaps by percentages of most common use, at best, after reading thousands of articles and adapting. I must teach the law first, then allow her to decide when the law is not wide enough. -But first she will be a child that listens exactly to what I teach . . .
Go to Top of Page

toborman
Hooked Member



USA
291 Posts

Posted - May 02 2009 :  05:32:03  Show Profile  Visit toborman's Homepage  Reply with Quote
Harry needs to be able to remember three types of memories: semantic, procedural, and episodic. He remembers and recalls all three to some extent. semantic in the knowledge, english, and inference tables, episodic in the conversation table, and procedural in individual procedure tables. even though the records in these tables are physicaly structured the same, this separation allows unique content and organization of the records.

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

toborman
Hooked Member



USA
291 Posts

Posted - May 03 2009 :  00:28:38  Show Profile  Visit toborman's Homepage  Reply with Quote
When Harry and Cortona meet for the first time, what will they say?

If we treat our VH like a traveler from a foreign country who has learned some basic vocabulary, what kind of conversation would we have? Here is an initial vocabulary for travelers in a foreign country. What skills and knowledge would our VH need in order to use these words in conversation? Whatís missing? Standard NLP chatbot approach or something more? Is there a basic set of vocabulary that all VHs should know?
∑ Yes
∑ No
∑ Thank you
∑ Thank you very much
∑ You're welcome
∑ Please
∑ Excuse me
∑ Hello
∑ Goodbye
∑ So long
∑ Good morning
∑ Good afternoon
∑ Good evening
∑ Good night
∑ I do not understand
∑ How do you say this in [English]?
∑ Do you speak ...
∑ English
∑ French
∑ German
∑ Spanish
∑ Chinese
∑ I
∑ We
∑ You (singular, familiar)
∑ You (singular, formal)
∑ You (plural)
∑ They
∑ What is your name?
∑ Nice to meet you.
∑ How are you?
∑ Good
∑ Bad
∑ So so
∑ Wife
∑ Husband
∑ Daughter
∑ Son
∑ Mother
∑ Father
∑ Friend
∑ Where is the bathroom?


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

Hatcheda
Curious Member



27 Posts

Posted - May 03 2009 :  05:03:33  Show Profile  Reply with Quote
From another thread I read . . . Somewhere in the pages of this, I think . . .
http://www.vrconsulting.it/vhf/topic.asp?whichpage=23&TOPIC_ID=906#17678
I like the idea of an:
Intro section.
Conversation section (refuses re intro)
Conclusion section.

Sounds like you are focused on 1 and 3. Which are good to focus on. I might try to group them under a category of each section.
At present, I am working on the 2nd part. I am working on a method of breaking down any sentence by identifying parts. My thought is to explode the string after running it through my grammar check. Then I will test each part against a list of Names, Verbs, Articles, Adjectives and such. I will cross check those with multi match against those with single match. Then decide the fate of the correct id of those with multi match.
I will then attempt to code my templates to recognize sentence part order. We will see . . . .
Go to Top of Page

Hatcheda
Curious Member



27 Posts

Posted - May 03 2009 :  21:26:50  Show Profile  Reply with Quote
Cortona can now break apart a sentence! :-) I will post some results later today . . .
Go to Top of Page

toborman
Hooked Member



USA
291 Posts

Posted - May 04 2009 :  02:50:54  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
Cortona can now break apart a sentence! :-) I will post some results later today . . .

Harry is anxious to see the results.

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

Hatcheda
Curious Member



27 Posts

Posted - May 04 2009 :  19:50:58  Show Profile  Reply with Quote
Sorry about the lack of post yesterday. -Fell asleep early.
Example of results from a quick test . . .
quote:

David is a Name
is is a Helping Verb
a is a Article
nice is a Adjective
tall is a Adjective
friendly is a Adjective
human is a Noun

It does always read is "a" and not "an" but that part is not long term coding. Just testing . . .
Code is listed below:
Grammar parts db and Cortona.au3 file hosted from my skydrive. Ėcut down Cortona to required parts.
Link below . . .
http://cid-dad9dc46d450d472.skydrive.live.com/browse.aspx/Cortona

Still requires much work. - I will need templates (fewer) and also some selection logic for multiple matches. -but a great step, I think!
Go to Top of Page

toborman
Hooked Member



USA
291 Posts

Posted - May 04 2009 :  22:23:06  Show Profile  Visit toborman's Homepage  Reply with Quote
quote:
David is a Name
is is a Helping Verb
a is a Article
nice is a Adjective
tall is a Adjective
friendly is a Adjective
human is a Noun

Harry says "nice work".

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

Hatcheda
Curious Member



27 Posts

Posted - May 05 2009 :  03:22:00  Show Profile  Reply with Quote
Thanks! Updated . . .
http://cid-dad9dc46d450d472.skydrive.live.com/browse.aspx/Cortona

Edited by - Hatcheda on May 06 2009 21:11:39
Go to Top of Page

TheMikh28
Curious Member



USA
39 Posts

Posted - May 05 2009 :  18:36:04  Show Profile  Reply with Quote
Anyway, I was checking out Grammar.au3.

This may or may not be of any help, but what I did with own my project (Fadela) was create two separate "sub-databases" (among the other sub-databases within the main database serving different purposes) that served a similar purpose (handling of singular/plural items, as well as terms that I wanted to be filtered to another.)

One database contains information for altering incoming text from the user. For instance, if I said something along the lines of "I am Michael," the phrase will be filtered to something like "Michael [is word] Michael" (Incoming "I" filtered to "Michael", for all practical purposes in my one-to-one interaction with the bot). Not only that, but the general sentence structure is stored for future application.

The second database contains information for altering outgoing text from the bot. For instance, if I somehow requested that the bot tell me my name, it will respond (applying past stored sentence structures) with something like "You [is word] Michael." In other words, "You" would be what "Michael" filters to in the case of outgoing text.

The use of databases is so that I don't have to go and alter the source code in order to manage these "filters." In addition to this, I can just request that the bot make the desired changes to the databases, so I save time and energy by not touching the code.

As for terms that are specified to be filtered *after* the terms have been used [ie, several times], I plan (in the near future) to write a process in which the bot runs through its database of terms and concepts, checking for inconsistencies, duplicates, and anything else that isn't right or needs to be modified, among other things; I call this process "Idling/Sleeping." This system should be able to effectively handle that problem.

[M*P]

Edited by - TheMikh28 on May 05 2009 18:44:57
Go to Top of Page

Hatcheda
Curious Member



27 Posts

Posted - May 06 2009 :  21:08:37  Show Profile  Reply with Quote
TheMikh28: -Thanks for the info.
About grammar.au3 -it's a work in process. Everyday it has changes, so if you see any issues, just keep checking or feel free to suggest changes.

Toborman:
I have made some great changes! Using sentence parts I am now able to use arrays and templates together to understand what the user is typing.

For Ex
David is a nice, tall, male, human while John is a short, mean, zesty, male, monkey.

Cortona is then able to store everything listed above.
I also added voice a while back. -be sure to check the console if you donít have the setup for voice or you can download it. (the speak function provides the info needed)

All of my files must be downloaded again as they have changed.

Next I plan to write something to best determine multi match for the sentence parts.

Download from:
http://cid-dad9dc46d450d472.skydrive.live.com/browse.aspx/Cortona

Comments are very welcome!

Go to Top of Page

toborman
Hooked Member



USA
291 Posts

Posted - May 07 2009 :  00:05:56  Show Profile  Visit toborman's Homepage
quote:
Toborman:
I have made some great changes! Using sentence parts I am now able to use arrays and templates together to understand what the user is typing.

For Ex
David is a nice, tall, male, human while John is a short, mean, zesty, male, monkey.

Cortona is then able to store everything listed above.

You're making good progress quickly.

More about Harry's design. Harry is command driven.
For Ex
"David is a nice, tall, male, human." is treated as
"remember David is a nice, tall, male, human." stores the descriptors
or "forget david is nice." removes that attribute from the db, and
"describe david." recalls David's descriptors.

These commands indicate the users intention. Harry responds to 20-30 commands.

http://mindmap.iwarp.com
Go to Top of Page
Page: of 25  Topic Next 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 0.13 seconds. Snitz Forums 2000