The AI Engine

I'm not sure what the rule is about double parentheses. All I know is that I've had some not work, so I avoid them.

Alright, odd question.

Now that you can make 'self' memories, I'd really like to start setting defaults for many of my bot's favorites. Problem is, I don't know what plugin to use for a lot of them.

Example: <?PF self:default "(favoritefood)" as "pizza" ?>

What exactly would I put for favorite food? Or other things?

For me, it seems like this works: I (hate|do not (really|) like|loathe) that tone

But this doesn't: ((adjartnounprep)|(adjartnounprep)s) (is|are) cool

Then again, I don't think I've ever had a keyphrase work with double parentheses in the beginning.

So, I tried this experiment.
We know that a keyphrase with a ( at the beginning will work.
We do not know if double parentheses will work in a structure with two of them consecutively.

Knowing that, I tried this keyphrase.

(gibberish|) ((adjartnounprep)|(adjartnounprep)s) (is|are) cool

Didn't work.
Therefore, this is my theory:

double parentheses found anywhere in a keyphrase will not work
A keyphrase will work if it has a similar structure to this:
I (like|love| appreciate (very|) much)

It works if it has a sentence structure like this.
I (love|(really|) like)

The last one worked in a seperate experiment using the keyphrase I (bob|(really|) bob), and it worked.

That's my story and I'm stickin' to it.

Cool, that's really useful! Thanks so much for posting that and sharing it with us.

Yeah, that's good to know! Thanks, alc003!

In addition to my other problem, I want to bring up one other issue that still nags me. I'm sure everyone's noticed that bots will utter one last phrase after declaring that they have to go, which results in some pretty awkward endings to convos. Are they ever going to be forced to stick to their guns and hang up immediatly?

You can make them stick to their guns by using seeks. Just add a seek to your xgottago and xgoodbye, and the awkward endings will stop.

Problem with...ahem...recognizing foul language.

So Frizella appears to have stopped recognizing "shit" as a keyphrase anymore. I was fooling around with her "obscenity filter" (the responses that deal with foul language) and she stopped recognizing "shit" entirely. I haven't tried any other swears (but I will!), and I've removed every last shred of AI script that was associated with "shit", and it is ranked at 50, for emotions from -5 to 5...but when I put "shit" through debut it comes through as blab, and I don't get any "shit" responses.

Ideas?

"Shit" seems to be one of those words that are not recognised when used alone (or isolated by commas, semi-colums etc.). See Bug Stomp posts 4893 to 4897 for other culprits.

Thanks, Patricia! Sure enough--the chatter has to give it a little more effort than just a single expletive. I missed those Bug Stomp reports.

Heh--I meant "debug", not "debut" in my original message. "Introducing...Shit!"

Say, is there any way to set a memory to be conditional upon a bot's like/dislike of a person? As in, the bot would only speak good/ill of a person depending on how many smileys are recorded for them?

Sure! Just set the emotional range to the right of the keyphrases high (say, between 3-5) for responses for people it likes, and low for people it doesn't. Be sure to have a response for every emotional range, or at least one all encompassing one, or you'll get "Sorry, there are no valid responses..."