Newcomers

I'm working on a fortune telling bot called "Madame von Scam". I want to keep it simple and clean; she is just supposed to interpret question *types* and give relevant answers.

Nonetheless, I want her to discern questions as "How can you read my future" and "How will I be able to get that job", and thought about "how * (i|me)" and "how * (you)". I did some testing and she doesn't seem able to make that distinction. The you-variation is always neglected, and it goes to xnone. Howcome?

examples-->

user: how * (i|me)
answer: A good way to do that would be without your (noun-qual-amp). Remember that you are a (noun-equal-pos).

user: how * (you)
answer: Temperature is important. And patience. I won't tell you all my secrets, but you have to be a bit of a (noun-equal-neg) to do this job.

Maybe you can change the KP from

how * (you)

to

how * you

LarsB's suggestion makes sense, for (i|me) is an ad-hoc plugin, whereas, when the machine sees "(you)", it looks for a stored plugin named "you".

Another way would be

how * (you|you)

(This would be better if you have a (key2) in a response.)

Damnit, the brackets of course... Eventually changed it to:

how * (you|your)

Thanks!

( RomulusKesher
what can you tell me about my life
Madame von Scam
Certainly not what you think now. It would you make stupid. Swallow some crushed rock from time to time.

lol)

The thing with such a keyphase as how * (you|your) is that it will pick up
How nice of you
How old is your mother
How pretty you look
how can you eat worms etc

It's just too broad.

I know, but as my bot is very field-specific, chances are relatively higher that the string will refer to her abilities as a coffee reader

using phrases like
how can you
how can i
how will you
how will i
when will you
when will i

will work better. then you can use responses like
You will not (postkey) in the near future.
so it will respond
l
when will I get a job?
You will not get a job in the near future.
when will I get married
you will not get married in the near future.

Thanks, prob123. Every question-word (which, what, how, why, when..) will indeed need its own semantic distribution. I'll do some experiments with the postkey feature.

I am just trying out the new templates on one of my bots. They are great. I would recommend them. If you don't want a keyphrase just delete it.

I notice that I have some keyphrases that might not make sense with out the responses, like
(starts|starting|beginning|begins) with (the letter|) (*)
what number comes after ([1234567890,]+) (re)
If anyone needs the rest let me know

Am I able to look at all the commands for someones bot?