AI Zone

RICH Learning examples

2013-05-20T08:26:13-06:00

Several people have tried to access RICH(s) learning processes recently and havent quite hit on the syntax, so I thought Id publish an example here in case anyeone cares to give it a try in the future. There are no special commands such as “Learn” its all handled through NLP. This is an automated learning sequence.

Teacher: hello
Vince: Hi
Teacher: do you know anything about lobster
Vince: No, but now that you’ve mentioned I’ll remember.
Teacher: a lobster is red in color
Vince: Well I know what A lobster is, but I didnt know it had red color. Thank’s, I’ll remember that.
Teacher: a lobster can also be blue in color
Vince: Well I know what A lobster is and I know that it could be other colors, but I didnt know it could also be blue.
Teacher: a lobster has 8 legs
Vince: Well I know what A lobster is, but I didnt know it had 8 legs. Thank’s, I’ll remember that.
Teacher: a lobster has eight legs
Vince: I believe your correct about that, at least in some cases.
Teacher: a lobster has legs
Vince: As far as I know your correct about that.
Teacher: a lobster has 2 claws
Vince: Well I know what A lobster is, but I didnt know it had 2 claws. Thank’s, I’ll remember that.
Teacher: a lobster has two claws
Vince: I believe your correct about that, at least in some cases.
Teacher: a lobster has claws
Vince: As far as I know your correct about that.
Teacher: a lobster has 2 eyes
Vince: Well I know what A lobster is, but I didnt know it had 2 eyes. Thank’s, I’ll remember that.
Teacher: a lobster has two eyes
Vince: I believe your correct about that, at least in some cases.
Teacher: a lobster has eye stalks
Vince: Well I know what A lobster is, but I didnt know it had eye stalks. Thank’s, I’ll remember that.
Teacher: a lobster has eyes
Vince: As far as I know your correct about that.
Teacher: a lobster has a tail
Vince: Well I know what A lobster is, but I didnt know it had 1 tail. Thank’s, I’ll remember that.
Teacher: does a lobster have legs
Vince: Well A lobster is probably born with legs although it does not necessarily have to have legs My memory May be A little fuzzy though.
Teacher: does a lobster have eight legs
Vince: Well in at least some cases A lobster has eight legs. My memory May be A little fuzzy though.
Teacher: does a lobster have claws
Vince: Well A lobster is probably born with claws although it does not necessarily have to have claws My memory May be A little fuzzy though.
Teacher: does a lobster have two claws
Vince: Well in at least some cases A lobster has two claws. My memory May be A little fuzzy though.
Teacher: what color is a lobster
Vince: Well, in at least some cases A lobster is red.
Teacher: is a lobster red
Vince: Well, in at least some cases A lobster is red.
Teacher: can a lobster be yellow in color
Vince: Well, I know what an lobster is, but I dont know if it can be yellow in color. You can try saying “A lobster is [something]”. If I know you, I might take your word for it.
Teacher: can a lobster be blue in color
Vince: Well, in at least some cases A lobster is blue.
Teacher: does a lobster have eyes
Vince: Well A lobster is probably born with eyes although it does not necessarily have to have eyes My memory May be A little fuzzy though.
Teacher: does a lobster have two eyes
Vince: Well in at least some cases A lobster has two eyes. My memory May be A little fuzzy though.
Teacher: does a lobster have a tail
Vince: Well A lobster is probably born with tail although it does not necessarily have to have tail My memory May be A little fuzzy though.
Teacher: goodbye
Vince: Yeah see you later

There are still a few things to iron out, as you can see it still has problems with singular to plural in creating an object. Its best just to make something singular (Lobster instead of Lobsters) Some of the replies could be more accurate, for instance “Well, in at least some cases A lobster is blue.” would probably better reflect the logic if it stated that variations on coloring might reflect portions of an object rather than the entire object. (Carrots are orange, carrot tops are green…entirely green carrots are…Dr. Seusish) Originating attributes can be anything but only size, shape, and color are recognized as secondary attributes. Thats just a matter of programming time. There are still problems differentiating between theoretical and actual objects. Sometimes it correctly makes the distinction;

What color is a carrot (actual object, actual request)
What color is a blue carrot (theoretical object, logical request)

So that statements such as “Fido is a dog” will probably be interpreted as “I have a dog named fido” which creates an object named Fido with you as the owner.

and permutations on syntax are not always picked up, although these are usually assimilated into the library within an hour or so.

“Hi dee ho there friends and neighbors what the [redacted] do you know about [redacted] lobsters”

would probably be missed at first

Objects are persisted, attributes that are found to be incorrect are overridden during a back propagation pass.

Thanks to everyone who visits, your input is always invaluable

Vince

Oh, and I have NO idea why it suddenly decided that every “A” in a sentence should be capitalized LOL

Sentinel - Cloud based Personal Assistant

2013-05-20T20:06:19-06:00

So my initial project that have been altered,changed and segmented into different phases have finally reached the final stage of being available for private beta.

Its a cloud based personal assistant currently only available for windows devices (desktop, tablet & phone).

What makes this smarter than other Assistants? Well for one, while Siri and the likes can perform one request at a time. Sentinel will be able to perform several requests at the same time. That means you can ask it to find nearby restaurants while also sending a text message to friend.

Enabling true natural conversation.

Another is history. It will be able to reminder what the topic of conversation is and what you last talked about.

Another is condition based reminders. For example asking it to remind you to call someone if the Heat loses tonight.

and much more

Does anyone here have windows phone 8? It would certainly be of great benefit if you can beta-test the application with extreme scrutiny.

Without further ado tonight i will show you all the Reverification you all be waiting for! lol

An example of a thinking machine?

2012-07-05T08:43:23-06:00

I thought I would break out this experiment as an opinion poll. . .

This is a very limited domain thought experiment, 1 input, 1 output to see what everyone “thinks”:

USER:What is the number between twenty one and twenty three?
AI: Twenty two. 

This is an actual input/output from Skynet-AI. The numbers 21, 23, and 22 do not exist anywhere in Skynet-AI’s programming or database. Neither do the words; “twenty one”, “twenty two” or “twenty three”. The AI writes its own code to understand the natural language input, solve the problem, and produce natural language output.

So let me ask you, in this example, does Skynet-AI “think”?

To give you a background in how people think, this may be of interest:
Tracking Children’s Mental States While Solving Algebra Equations

Harumi

2010-12-26T09:39:06-06:00

Hi, my bot Harumi is already available on Chatbots.org but I wanted to create a thread for it in order to have more feedback on it.
My chatbot must be downloaded to be used. You will need open office; internet is not necessary (except for downloading!).
Harumi is a french chatbot able to learn from discussions. This is the feature lists :

- Calculate operations with + - * / and ( )
- Record appointments
- Record multi alarms, on time launching a website or a file.
- Mood is symbolized by a number between 0 and 100. Smileys, considered as feelings can increase or decrease mood.
- Apparence of Harumi change during the time of the day.
- Harumi plays chess and tell you what’s going on during the game.
- when typing a sentence, a proposition eventually pops up next to the entry window. Press F1 to validate this proposition.
- Record and reming you things to do
- Randomly reming you informations she learned
- Linking 1000 answers possibles for seach stimulus. 65000 stimulus possibles. When a question is asked regarding a stimulus and many answers are possible, Harumi search for the best answer possible.
- Stimulus can be edited with boolean logic : “Jeff AND phone”, “[Jeff OR my master] and phone”
- Database of synonyms crossed with user request.
- Everything is editable : sounds, images, stimulus…
- Giving IP adress and EMAC adress
- Dynamic detection of internet and battery.
- Recognizing similar images (very close one actually, taken with a bit different viewpoint).

Ontology

2013-05-17T05:42:29-06:00

Hello guys,

I am trying to define my story domain but I do not know how to do it ?? can any one help me and tell me how I can specify the story domain for my system (the ontology) ?? I mean is there any applications for that ?

Cheers

ChatScript to email server

2013-05-11T04:29:37-06:00

Anyone tied ChatScript to an email server yet ?

The object would be to offer up users of the system a copy of their most recent transcript, or content/details that would be useful to their use of the system more broadly.

New syntax overhaul for RiveScript?

2012-02-15T11:21:07-06:00

I’ve spent much more time writing RiveScript than I have writing in RiveScript, and I’ve been thinking for a while about overhauling what the syntax looks like for writing the code. For example, some of the limitations in RiveScript 2.0’s current syntax are:

Space is kind of cramped. If you want a conditional that results in random replies, you have to write some ugly code to do it.
Conditionals can’t stack. If you want to check multiple variables, one condition needs to redirect to another trigger for the next check. You can be clever and use topics to “hide” the private trigger, but it’s messy.

Compare this syntax:

+ my name is *
* <get name> == <formal> => I know, you've\\s
   ^ told me your name before.
* <get name> == <bot name> => <set name=<formal>>{random}
   ^ Wow, we have the same name!|
   ^ What a coincidence, that's my name too!|
   ^ What are the odds that wed have the same name?
   ^ {/random}
- <set name=<formal>>Nice to meet you, <get name>! 

With what it could look like:

trig my name is *
    if (<get name> == <formal>) {
        reply I know, you've told me your name before.
    }
    elsif (<get name> == <bot name>) {
        set name = <formal>
        random {
            Wow, we have the same name!
            What a coincidence, that's my name too!
            What are the odds that wed have the same name?
        }
    }
    else {
        set name = <formal>
        reply Nice to meet you, <get name>!
    } 

I’m open to thoughts/suggestions about this. The new syntax wouldn’t be backwards compatible (it’d be RS 3.0), but since RS is so trivial to parse, an upgrader program for old scripts would be easy to make. Here’s my full mock-up of the new syntax idea:

// RiveScript 3.0 syntax mockup

! version = 3.0

// Same syntax as RS 2.0
! array colors = red green blue yellow
! sub who's = who is
//'

// The new version of the > begin block
begin {
  // New version of conditionals
  if (<get name> == undefined) {
    // New way of setting vars, the <set> tag is removed
    set topic = newuser
  }

  if (<bot mood> == normal) {
    // The reply commands stack, more than one reply = they concatenate.
    reply <ok>
  }
  elsif (<bot mood> == angry) {
    // No more difference between {tags} and <tags>, we use <tags> everywhere
    reply <uppercase><ok></uppercase>
  }
  elsif (<bot mood> == sad) {
    reply <lowercase><ok></lowercase>
  }
}

topic newuser {
  // Maybe + will be an alias for trig? And - an alias for reply?
  // Triggers make "implied" blocks (no need for { and }). The block ends
  // when a new trigger is found or a parent block ends (a } is found).
  trig *
    set topic = newuser_1
    reply Hello, I don't know you! What's your name?
}

topic newuser_1 {
  // Wildcards are the same: _ for words, # for numbers, * for anything
  trig my name is _
    set name = <formal>
    set topic = newuser_2
    reply Nice to meet you, how old are you?

  // 'alias' is the new @.
  trig _
    alias my name is <star>

  trig *
    reply I'm sorry, what was your name?
//'
}

topic newuser_2 {
  trig i am # years old
    set age = <star>
    set topic = random
    reply Ok, I will remember you are <get age> years old.

  trig #
    alias i am <star> years old

  trig *
    reply I'm sorry, how old are you?
}

// The indentations are optional but recommended for style
trig hello bot
reply Hello human!

// aliases can stack with replies.
trig * or something
  reply Or something.
  alias <star>

// Random replies. To continue a line, end it with a \\ character.
// The random keyword stacks with reply too.
trig how are you
  random {
    I'm great, how are you?
    Good, you?
    I'm doing well, how about you?
    I am doing so well you \\
      wouldn't even believe it.
//'
  }

// A reply stacking example. Replies are concatenated using space characters
// now (this may be a configurable option which may be dynamically changed)
trig tell me about linux
  // Begin a reply.
  reply Linux is a free operating system.

  // Set some variables.
  set topic = linux

  // Continue.
  reply It was written by Linus Torvalds and first released in 1991. 
    One example of a distribution of Linux is

  // A bit of random
  random {
    Fedora.
    Ubuntu.
    Gentoo.
    Mandriva.
  }

  // Throw in a conditional.
  if (<get name> == Linus) {
    reply You have the same name as the creator of Linux!
  }

// You can redefine global variables just as easily now too.
trig turn debugging (on|off)
  // Only the botmaster!
  if (<id> == <bot master>) {
    if (<star> == on) {
      // Variables are lexically scoped to the current request.
      var debug = true
    }
    else {
      var debug = false
    }

    // Set the debug mode.
    ! global debug = <var debug>

    reply Debugging has been turned <star>.
  }
  else {
    reply You are not the botmaster. I won't listen to you.
    //'
  }

// Object example.
object md5_encode perl {
  my ($rs, @args) = @_;

  use Digest::MD5 qw(md5_hex);
  my $hash = md5_hex(join(" ", @args));

  return $hash;
}

// Call the object example.
trig what is the md5 hash of *
  // If the character(s) used to concatenate reply segments is configurable,
  // it could be overridden on a per-reply basis. Here we could set it to
  // be a blank string. BTW, quotes can be optionally used on all "set" like
  // commands, but are optional. They're handy for setting empty strings or
  // spaces as values though.
  option concat = ""

  reply The MD5 hash of "<star>" is "
  call md5_encode <star>
  reply ". 

Neobot and Novabot v0.7a will soon be released

2013-05-14T17:02:56-06:00

Neobot and Novabot v0.7a will soon be released May 20, 2013. Neobot is already available, but Novabot (female counterpart of Neobot) will be released for the first time on Launchpad.

https://launchpad.net/neobot

How could Xaiml be improved?

2013-05-14T07:17:20-06:00

NOTE: If you are not familiar with Xaiml, then you may want to read this https://launchpadlibrarian.net/133079399/Xaiml.TXT.

How could Xaiml be improved? What tags do you feel that it should have? What could be done to make it the best chatbot knowledge database (KB)?

What actions should be taken to make Xaiml popular?

2013-05-14T07:21:50-06:00

Xaiml Documentation and specification: https://launchpadlibrarian.net/133079399/Xaiml.TXT

What actions should be taken to make Xaiml a popular chatbot database markup language?