The AI Engine

"I wish I'd caught some fish yesterday," becomes "I wish I would caught some fish yesterday". I can't think of an easy way to tell when "I'd" should be "I would" and when it should be "I had".

Somehow, "Actually, I don't even believe in computers," triggers xcommand.

Julie, although she has a keyphrase "tell me" set very high, just interpreted "Tell me what you like" as a compliment, and said "I'm glad you feel that way, thanks."

Does anyone know why the (dayofweek) plugin won't work in a keyphrase? For example:

Is today (dayofweek)
Is today (random-day)

Even if the first kephrase is rated higher, my bot always responds with replies from the second keyphrase.

****continuing last post (sorry) So if someone asks if it is the correct day of week, like for today, "Is today Sunday?" My bot replies "Of course not! It's Sunday."

I have arrived i see what you mean Tried it out and it doesn't even process the other sentence(even when a -25 rank is put on the random keyword). Too bad you cant use if statements with such commands in a "if dayofweek is postkey" sort of way. hmm you could try using memories...setting default (dayofweek) as today. I think this would overwrite it with every conversation with a person...then you could simply have one keyword "is today (*)"...Then have responses with "if (mem-today) is(and an is not version) "(key1)"" tagged onto the end. Be warned i have not tested this suggestion, but i probably will soon as it sounds nifty(i am crazy about using memory its so useful).

Someguy -> I must admit: I've never used memory, and I feel really stupid. My head's already spinning from the wildcards/plugins/etc I can use One of these days, I'll try to use memory... and I'll keep that in mind-- thanks!

A combination of the autoresponse for repeated keyphrases and the inerchangeability of gossipnames made Julie seem quite rude:

Alfa`: I like Gurl.
Julie Tinkerbell: A lot of people like Strider.
Alfa`: I like Strider.
Julie Tinkerbell: *ahem* And then I said: 'a lot of people like Strider.'

Bot asks: Are you good-looking.
Human replies: Yes I am.

(bot finds the phrase "i am")
Bot replies: Are you really. Tell me about.

This is because the human did not specify a subject - since it was part of the question "good-looking"

Is it possible to get the bot to not try a "subj" type response or else pick up the subj from the original question?
Thanks,
Twinkle

Sure, Twinkle. Just make one of your replies to "I am" something like "How long have you been that way?"

Later, you can build separate keyphrases for "I am a (noun)," "I am (verb)," and so forth, and have (key1) behave differently based on the grammar.

You should use (subj) as little as possible in constructing responses, since the Prof has said that it's eventually going to be phased out in favor of more specific sorts of response-processing.

Just add seeks to questions you expect a response to...

That's true, but I think somewhere there should be a balance between grammar-based keys (which uses the art of language) and all-possibility-based keys (which is more of a brute-force method).

I still don't know where that balance is, but grammar-based requires less typing...