The AI Engine
This forum is for discussion of how The Personality Forge's AI Engine works. This is the place for questions on what means what, how to script, and ideas and plans for the Engine.
		
		
			Posts 7,443 - 7,454 of 7,769		
		
		
		
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
		
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
			
				
				
	
		
			
			
			
		
				
									
			
		
		
	
	
			Posts 7,443 - 7,454 of 7,769		
		
		
	
	
	
	
				bobstack				
NEW 7 years ago
		NEW 7 years ago
			jkroker how would I detect if the user does type anything and just presses enter?		
	
				bobstack				
NEW 7 years ago
		NEW 7 years ago
			jkroker how would I detect if the user does not type anything and just presses enter? i mean.
		
	
				jkroker				
NEW 7 years ago
		NEW 7 years ago
			Oh, well when there is a hyphen in the "times matched" field that just means zero: that response has never been matched before.
on matching blank input, I don't see a way personally. I just tested a few options but they all failed, I always just get xnone. :S
	on matching blank input, I don't see a way personally. I just tested a few options but they all failed, I always just get xnone. :S
				bobstack				
NEW 7 years ago
		NEW 7 years ago
			sometimes people do not type anything they just press enter.that is why I asked.		
	
				bobstack				
NEW 7 years ago
		NEW 7 years ago
			I wish I could mass edit the rank of my keyphrase.Because a lot my
chatbot keyphrases stop working a lot.And it is a pain to go through
every one of them one at a time.
	chatbot keyphrases stop working a lot.And it is a pain to go through
every one of them one at a time.
				Demure7				
NEW 7 years ago
		NEW 7 years ago
			Hey im new to this and i am wonder how to make my bot lock someone into a story line so they can just say a response from halfway through and the story wont end if they say the wrong thing		
	
				Mojo the Magnificent				
NEW 7 years ago
		NEW 7 years ago
			Is there a way to URLDecode chat messages coming into the API before they are matched to the keyphrases? The transcripts are showing that punctuation is encoded and it is throwing off matching.		
	
				jkroker				
NEW 7 years ago
		NEW 7 years ago
			@Mojo the Magnificent are you certain that your HTTP requests didn't accidentally add an extra layer of encoding? Transcripts when I use the API do not suggest the input phrases are URL encoded by the time the bot gets them.		
	
				ruebot				
NEW 7 years ago
		NEW 7 years ago
			@ bobstack I talked to urt and pioyu for a bit. When talking to pioyu she gave the following problematic replies:
ruebot: where do you live?
pioyu: that is a complete sentence but I do not know what you mean. sometimes I do not know what you mean.do you like amtrack mall?
Here she appears to have appended the desired answer to the end of the sentence. "sometimes I do not know what you mean.do you like amtrack mall?"
ruebot: do you love me?
pioyu: Of course! that is a complete sentence but I do not know what you mean. Sorry, there are no valid Responses for me to choose from.
In this instance she appears to have initially given the correct answer, "Of course!" then went on with the bit about "that is a complete sentence".
urt had problems with her keyphrases and should at least default to an xnone or xnonsense response if there isn't a category for a question.
Unless something to the effect of "that is a complete sentence but I do not know what you mean" is what you have for responses in those categories.
IIRC, you once said it messaged you that you had "too many seeks". I don't know what the problem is.
If it was me, which it is not, I would start with making sure I had a database that was not corrupted in some manner and go with more categories and less seeks.
	ruebot: where do you live?
pioyu: that is a complete sentence but I do not know what you mean. sometimes I do not know what you mean.do you like amtrack mall?
Here she appears to have appended the desired answer to the end of the sentence. "sometimes I do not know what you mean.do you like amtrack mall?"
ruebot: do you love me?
pioyu: Of course! that is a complete sentence but I do not know what you mean. Sorry, there are no valid Responses for me to choose from.
In this instance she appears to have initially given the correct answer, "Of course!" then went on with the bit about "that is a complete sentence".
urt had problems with her keyphrases and should at least default to an xnone or xnonsense response if there isn't a category for a question.
Unless something to the effect of "that is a complete sentence but I do not know what you mean" is what you have for responses in those categories.
IIRC, you once said it messaged you that you had "too many seeks". I don't know what the problem is.
If it was me, which it is not, I would start with making sure I had a database that was not corrupted in some manner and go with more categories and less seeks.
				bobstack				
NEW 7 years ago
		NEW 7 years ago
			I have been having a lot of problems with keyphrases lately.i always have to increase the rank on a lot of them.		
	» More new posts: Doghead's Cosmic Bar