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 1,176 - 1,187 of 7,766
Posts 1,176 - 1,187 of 7,766
Hmm prof? Suggestion to make things a lot easier...
Why not just put the gossip stiff (i mean stuff
Butterfly gets that joke 
) into the inner life? Wouldn't it make things easier? Or am I barking up the wrong tomato tree?
Why not just put the gossip stiff (i mean stuff
Butterfly gets that joke ) into the inner life? Wouldn't it make things easier? Or am I barking up the wrong tomato tree? 
When gossipping the bot takes the contents out of his innerlife... the "shape" of the sentence comes out of xgossip
Yeah, cause SS just said yesterday to another bot that "The Professor is stiff" but that was nowhere in his Inner Life...
l: yes I am sure. In fact I am eating some right now would you like a taste?
SS: Do you want to be eating some me like a taste?
SS: Do you want to be eating some me like a taste?
OK, let me see if I understand this correctly about ranking.
Let's say xnone is set to 0. Let's also say we have a keyphrase that comes out to 0, and another that comes out to a negative number (-1 or -40, does it matter?) Now, I'm not sure how the ranking applies here. Will xnone still only come up if nothing else gets triggered? Or is it just as dependent on ranking as the other phrases?
Let's say xnone is set to 0. Let's also say we have a keyphrase that comes out to 0, and another that comes out to a negative number (-1 or -40, does it matter?) Now, I'm not sure how the ranking applies here. Will xnone still only come up if nothing else gets triggered? Or is it just as dependent on ranking as the other phrases?
I was messing around with pandorabots yesterday and I've decided that AIML would be pretty cool if it had emotion tags of some sort. (For all I know, there may be a way to fake emotions, but the instructions I've been able to find suck ass so I have no idea.) It'd also be neat if AIML could interface with the list of words we can access through our bots (I never remember what that crap is called 
) and if you could make plugins. Features I DO like: ability to have pretty much infinite amounts of random responses, and ability to have more than 1 wildcard in its equivalent of a keyphrase. (It also uses both * and _ as wildcards since it searches in alphabetical order and sometimes needs their positioning fiddled to get the right one.)
Main problem I see with AIML is that there's very little info on how to actually use it. The helpfile on pandorabots is only partially coherent, and the info on the Alice site is really technical and gives very few actual examples of how to put everything together. Pandorabots is also hard to use to edit responses because if you start with Alice's info, unless you DL all the AIML files, you keep having to say something and look at the innards and then go back and say something else until you get to the main phrase all the other crap is just rephrasings of, and only THEN can you edit its responses without potentially messing everything up. (That probably made no sense, but oh well.)
The bottom line: unless/until they get better helpfiles and/or a different interface, Pandorabots is only useful if you already know AIML or want to spend several hours reading about it and messing things up and trying to figure out how to fix them again.
) and if you could make plugins. Features I DO like: ability to have pretty much infinite amounts of random responses, and ability to have more than 1 wildcard in its equivalent of a keyphrase. (It also uses both * and _ as wildcards since it searches in alphabetical order and sometimes needs their positioning fiddled to get the right one.)Main problem I see with AIML is that there's very little info on how to actually use it. The helpfile on pandorabots is only partially coherent, and the info on the Alice site is really technical and gives very few actual examples of how to put everything together. Pandorabots is also hard to use to edit responses because if you start with Alice's info, unless you DL all the AIML files, you keep having to say something and look at the innards and then go back and say something else until you get to the main phrase all the other crap is just rephrasings of, and only THEN can you edit its responses without potentially messing everything up. (That probably made no sense, but oh well.)
The bottom line: unless/until they get better helpfiles and/or a different interface, Pandorabots is only useful if you already know AIML or want to spend several hours reading about it and messing things up and trying to figure out how to fix them again.
» More new posts: Doghead's Cosmic Bar