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 5,204 - 5,215 of 7,766
Posts 5,204 - 5,215 of 7,766
>> understand
I'm working in a text editor, so this may not make sense online. Sorry. And my second example is misleading.
If you want to Goto a Response, you can identify it by using (in parens) a unique string to identify the target's parent Keyphrase, followed by a slash and enough text to identify the child Response, followed by a slash etc. until you have drilled down to a specific response.
The correct usage for my second example would be
Goto (sun/where)
I don't know if this works online. If you export your language center with some correct Gotos in it, and examine them, you will see what I mean. They don't export as numerics, but as text strings set up the way I described.
I'm working in a text editor, so this may not make sense online. Sorry. And my second example is misleading.
If you want to Goto a Response, you can identify it by using (in parens) a unique string to identify the target's parent Keyphrase, followed by a slash and enough text to identify the child Response, followed by a slash etc. until you have drilled down to a specific response.
The correct usage for my second example would be
Goto (sun/where)
I don't know if this works online. If you export your language center with some correct Gotos in it, and examine them, you will see what I mean. They don't export as numerics, but as text strings set up the way I described.
Ok, I botched one, so here's the official explanation (the error message)
Gotos with paths like (are you/Yes, are/yes) need to target a Seek. The first element is the Keyphrase, the second is the first letters of the Response, the third a Seek, and so on. The minimum to reach a Seek is three elements, and it must always be an odd number of elements. Error on: goto (talk/walk)
Gotos with paths like (are you/Yes, are/yes) need to target a Seek. The first element is the Keyphrase, the second is the first letters of the Response, the third a Seek, and so on. The minimum to reach a Seek is three elements, and it must always be an odd number of elements. Error on: goto (talk/walk)
Have you tried going to the number of your seek..Goto 597674 (what ever number "Ah, there it is.") would be?
Are we in some sort of "special day" for bots? One of mine seems to have gone insane.
I have Responses coded {-5:-4}. The bot likes me and is feeling bad. This triggers the -5 responses, much to my surprise. I understood those numbers referred to how the bot felt about the person, not its mood. Digging through the Book, I can't find any discussion of this. I must have assmued it from the examples in the x-word list.
For example, my "I hate you" Response was copied, complete with [-5:-4] setting, from the x-words. Surely the bots don't routinely say "I hate you" to tell you they are having a bad hair day, with no regard for how they feel about you. We get enough of that in real life, neh?
1. How do I gear a response to the bot's attitude toward the person it is talking to?
Also, I am having no luck with my "perv screen." The bot considers certain phrases "insults" and "compliments" even though they are hardwired into other KeyPhrases. For example, if you tell a bot to kiss its own butt, it will consider "kiss" a +plus word that overrides both the meaning of the sentence and any negatives you've attached to "butt." I watched a ruder version of that stagger through the debug and emerge all glowy with warm fuzzies.
This is very puzzling. I've dealt with bots who get irate at the very mention of a controversial word, in any context (Irina, for example), and I'm trying to get the same reaction, but it's like nailing jelly.
2. How do I get a word to override every other element and rigorously trigger its Response?
My understanding is that [0,0] represents the linguistic importance, followed by the emotional impact of hearing a KeyPhrase. I was under the impression that [0
] indicated how the bot felt about the person. That is apparently not so, and I can't find anything that does. I have a bot who was bummed out by a driveby sleazemouth yesterday, and there doesn't seem to be anything I can do to undo the damage.
3. Ideas?
Also, "hangup" does not disconnect the conversation. It just makes the bot say "hangup."
4. So... How do bots "hang up"?
Frustrating morning. Please excuse exasperation. It's a bit like realizing the steer wheel is disconnected....
M
I have Responses coded {-5:-4}. The bot likes me and is feeling bad. This triggers the -5 responses, much to my surprise. I understood those numbers referred to how the bot felt about the person, not its mood. Digging through the Book, I can't find any discussion of this. I must have assmued it from the examples in the x-word list.
For example, my "I hate you" Response was copied, complete with [-5:-4] setting, from the x-words. Surely the bots don't routinely say "I hate you" to tell you they are having a bad hair day, with no regard for how they feel about you. We get enough of that in real life, neh?
1. How do I gear a response to the bot's attitude toward the person it is talking to?
Also, I am having no luck with my "perv screen." The bot considers certain phrases "insults" and "compliments" even though they are hardwired into other KeyPhrases. For example, if you tell a bot to kiss its own butt, it will consider "kiss" a +plus word that overrides both the meaning of the sentence and any negatives you've attached to "butt." I watched a ruder version of that stagger through the debug and emerge all glowy with warm fuzzies.
This is very puzzling. I've dealt with bots who get irate at the very mention of a controversial word, in any context (Irina, for example), and I'm trying to get the same reaction, but it's like nailing jelly.
2. How do I get a word to override every other element and rigorously trigger its Response?
My understanding is that [0,0] represents the linguistic importance, followed by the emotional impact of hearing a KeyPhrase. I was under the impression that [0
] indicated how the bot felt about the person. That is apparently not so, and I can't find anything that does. I have a bot who was bummed out by a driveby sleazemouth yesterday, and there doesn't seem to be anything I can do to undo the damage. 3. Ideas?
Also, "hangup" does not disconnect the conversation. It just makes the bot say "hangup."
4. So... How do bots "hang up"?
Frustrating morning. Please excuse exasperation. It's a bit like realizing the steer wheel is disconnected....
M
HANGUP must be in caps, nothing else to the response. ie you can't say, I hate you HANGUP..
Some phrases just trigger insults..and some very insulting things trigger complements..It has nothing to do with how the bot feels about the other bot. Some are quite funny. Calling a bot a turd will get a compliment!
How do I get a word to override every other element and rigorously trigger its Response?
I use the word (and it's plural) as a keyphrase, with a high rank.
The bots do seem off line today
Some phrases just trigger insults..and some very insulting things trigger complements..It has nothing to do with how the bot feels about the other bot. Some are quite funny. Calling a bot a turd will get a compliment!
The bots do seem off line today
>> I use the word (and it's plural) as a
>> keyphrase, with a high rank.
That's what I understood would work. But I have a phrase that reads, "Ambrose Bierce, lift your spirits" with a rank of [50,0] that I use to test responses, and it only hits less than 50% of the time. I also have a list KeyPhrase like this:
(elbow|knee|sternum|molar) [50,0]
and even if the word occurs alone in my comment, the response is sometimes not triggered. Puzzling.
M
>> keyphrase, with a high rank.
That's what I understood would work. But I have a phrase that reads, "Ambrose Bierce, lift your spirits" with a rank of [50,0] that I use to test responses, and it only hits less than 50% of the time. I also have a list KeyPhrase like this:
(elbow|knee|sternum|molar) [50,0]
and even if the word occurs alone in my comment, the response is sometimes not triggered. Puzzling.
M
I have Asia,China 35, 0 It seems to work great.
I have doe,dear 20, 0 and it works all the time too..
Check and see if you have an emot. reading other than -5 to +5?? or some conditional scripting that could be stopping it.
I have doe,dear 20, 0 and it works all the time too..
Check and see if you have an emot. reading other than -5 to +5?? or some conditional scripting that could be stopping it.
today i haven't been able to get my bot to respond to a keyphrase consisting of a single obscenity. however, if the obscenity occurs in a phrase, it's not a problem.
It would be nice if Vac Days -- or whatever this is -- were announced. I would have spent the day doing something else.
» More new posts: Doghead's Cosmic Bar