AI Zone Admin Forum Add your forum

NEWS: Chatbots.org survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..

CS 9.0 released
 
 

Now entering its ninth year, this version directly supports patterns representing fundamental meaning. Fundamental meaning consists of an actor, an action, and an optional actee.
In the active voice sentence “I love you”, the actor is “I”, the action is “love”,
and the actee is “you”. In the passive voice sentence “I was arrested”, there is no actor,
the verb is “arrested”, and the actee is “I”. Whereas in the passive voice sentence
“I was arrested by the police”, the actor is “police”.

Fundamental meaning patterns always have a verb, which as a keyword is designated as
“|arrest|” or whatever word or concept you want to detect.
A pattern which includes a fundamental actor is shown as
“~pronoun|arrest|”. One that includes an actee is
“|arrest|~police”, whereas one that has both actor and actee is
“~pronoun|arrest|~police”.

 

 
  [ # 1 ]

This set of features will prove to be very valuable to an already fully featured chatbot framework. 
It is great to see it is being actively updated by the original creator !

 

 
  [ # 2 ]

That seems like a very useful way to break down a sentence! I was playing with it in ChatScript 9 and perhaps I’m not understanding how it fits into a rule’s pattern:

e.g. I can’t get the following rules to trigger based on either of the inputs “cat eat dog” or “dog eat cat”
u: (cat|eat|dog) Response1
u: (“cat|eat|dog”) Response2

Can you please give an example of how I can incorporate fundamental meaning patterns into a rule?

Thanks!

 

 
  [ # 3 ]

Perhaps, the pronoun “it”, in place of “cat”.

~it|eat|~dog

 

 
  [ # 4 ]

Very well done Bruce. I believe in technical jargon this is called “semantic parsing”.

 

 
  [ # 5 ]

Technical jargon makes for lousy marketing jargon since almost no one understands what it means. But the average Joe can easily digest “fundamental meaning”

 

 
  [ # 6 ]

The bug is that I forgot to enable it for use directly in patterns. It currently only works in concept sets (where I tested it). I will fix that shortly (currently on vacation).

 

 
  [ # 7 ]

I agree. I always found it ironic that scientists who are trying to make computers understand language use language that nobody else understands.

 

 
  [ # 8 ]

Ah makes sense, thanks Bruce!

I tested the following, note that it appears to be the reverse order from the original post’s description - i.e. it seemed to be described there as using the order Object|verb|Subject while my test seemed to work only with Subject|Verb|Object order:

concept: ~policetest [I|arrest|police]
concept
: ~friendtest [friend|think|it]

topic
: ~topictest keep repeat (police)

u: (~policetestYou arrest police?
u: (~friendtestThey do, do they

Result:

alan: > Sometimes I have to arrest a police officer for violating law enforcement policies.
IVANOV:  You arrest police?
alan: > My friend thinks it funny when I arrest police.
IVANOV:  They do, do they

Something for when you get back from vacation: smile
Is there anyway to see the fundamental meaning for a given input - e.g. using a command like :tokenize?  That’d be super helpful to see how various inputs are boiled down into fundamental meaning terms.

Thanks!

 

 
  [ # 9 ]

I"think” the original post described the subject of the sentence in active voice as appearing to the left of the verb. And the subject of the sentence in passive voice appearing to the right of the verb.  Ie actor,action, actee order.

There is no way to see the fundamanetal meaning for a given input GIVEN that you are allowed to decscribe them as concepts for each position and there are lots of concepts for any word (for most things).  You can do :prepare and that will tell you the main subject, main verb, main object, and whether the sentence is PASSIVE or not. And for fundamental meanings you have described, it will tell you matching fundamental verbs, subjects, and objects.

 

 
  [ # 10 ]

fundamental meaning in patterns fix posted in ChatScript git

 

 
  login or register to react