The AI Engine

really?! well, that's a relief... I've only been stressing my heiny over the rankings. I think it's because once, I rated "you are" and "you are not" the same, and said "you are not being nice" and my bot responded with one of the "you are" keyphrases

i have arrived- I have asked myself the same questions, and feel your pain. lol
Re-ranking is one of the more time-consuming of the tweaking processes. Every time I add a new keyphrase, I have to be conscious of every single word contained within, and check the ranks of every keyphrase that might be caught to make sure I don't throw the whole thing out of balance...At one point I decided that I should really have a big wad of my keyphrases ranked lower, and I still haven't finished changing them- that was weeks ago...lol I would LOVE if I could somehow select and add or subtract, as you suggested, a certain # from all of them at once...in the meantime, I've started in the a's, and I'm to about the i's...I'll probably finish changing them all by sometime next fall...

Well, if you're using a text-editing program to work off-line, you could probably have it find and change all instances of a given number... Every "[8," could become a "[16," every "[7," could become a "[14," and so on ... though you'd want to "babysit" that process so as not to change anything you didn't want altered.

I don't think simply adding the same number to everything would help, because it's the difference between the ranks that would cause one to be taken instead of another -- you need to create more space between them, not just move them all higher up the number line.

no, not all, just "a wad" to bring down the average. Everythiing works, it's just more spread out than I would prefer it to be. I thought of using "find" in Word to edit the numbers,too, and that's a good idea, but I don't trust myself to not miss something and really screw the pooch, and plus I'm too chicken to use import (at all) yet. So far, I do all updating online, and just export my bot to save her "just in case".

I don't know if this is such a good idea, but I'm just going to keep my original rankings, and look at my bot's transcript to see if any of the present rankings should be changed. I don't think there's going to be THAT big of a problem with it, and it would save me from a lot of stress

Can we use the (dayofweek) plugin in a keyphrase? I had the keyphrase "is today (dayofweek)" and obviously, the answer would be yes. Then I made another one with a lower ranking: "is today (random-day)" and the answer would be no. However, every time I would ask if it was the correct day of week, it would go to the random day keyphrase and say NO. ???

Professor, thank you very much for the local plugin feature. I really (like|love|appreciate) that feature (tons|a lot|a whole lot).

One question - is it (more efficient|better) to use them in a keyphrase, or to use the comma-separated list, or does it make any difference?

example:

(did|does) (he|she|it)
vs.
did he, did she, did it, does he, does she, does it

I pretty much never use rankings.

Question: I was reading over my bot's transcripts and I noticed that if the other bot said "huh" or something short like that, my bot would respond to his own previous response. For example:

Battling Reality: Umm... is that a good thing?
mark the hunk: Huh?
Battling Reality: Yeah, that is a good thing. How did you know?

It's like he's talking to himself! he's not schizophrenic, I promise!!

Here's an odd one:

I gave Simulo a keyphrase for "who is ezzer" and ranked it fairly high. Then I asked him "Who is ezzer?" and he interpreted it as "possible gossip" (according to debug mode) ... then replied with x-emote-positive-very, because he really likes ezzer. (Awww...)

Of course, most of the x-emote replies have "you" in brackets, so he responded to the question by complimenting me.

Should this particular question be flagged as gossip? Is there any sort of workaround that would make it possible for him to answer the question?

Raise the ranking. I have specific keyphrases like that too, and they seem to work.

If I use the wildcard * and in a certain scenario it represents an empty space, and I've programmed the wildcard to be (key2) in a keyphrase, shouldn't the bot just make the (key2) part of the keyphrase blank?

If that didn't make any sense, here's an example of what I mean:

keyphrase: Do you * like *
response: Why? (prekey) do YOU like (key2)?

and someone talking to my bot says: what do you like?

Since (key2) is blank, shouldn't my bot's final response be: Why? what do YOU like?

Instead, he's saying: Why? what do YOU like (key2)?

Butterfly, I think it's this *specific* question ("Who is (friend)?") that's triggering gossip. Do you have any keyphrases for "who is X" where X is a specific name?

I don't mind it being a preprocessed question, but at least it should pull up mem-youare instead of x-emote.