The AI Engine

Hmmm, that's interesting. I can't seem to make it work either.

The key phrase "Midsummer night's" never gets called.
The key phrase "Midsummer night" never get's called.
The keyphrase "Middsummer" does get called, but only if he's given just "midsummer." Given "Midsummer Night" even with all three keyphrases above, xnone always gets called.

In preprocessing it seems to get turned into "Midsummer_Night," but a keyphrase for that never gets called, either.

Either I don't understand regular expressions (possible), or even they aren't helping me with this. Heck, I can't even build a reg exp that will catch "Midsummer," which seems very weird. I will keep trying.

I take it back. "Midsummer (re)" will catch "Midsummer," but it will not catch "Midsummer Night's" (even with nothing else in the sentence). This is just as far as Turing's Dad got.

Is it some sort of trouble with the capital letters?

Should we still do the [you] thing, or is it being phased out? I finally started putting it on one of my bots but then noticed the brackets were visible.

Hmm. [you] should still work. It only works in xemotes. Is that where you're using it?

Catching up a bit.. Incubus, the dynamic memory names seems like a good idea at first, but I think the number of possibilities would be overwhelming, since you wouldnt know the name of the memory and would have to test for every possibility.

I made a little progress with the Expectation system. It's rather rudimentary, and shows what most likely follows any given word, and can generate random sentences using that knowledge. The sentences are not at all inflected, so it's a little mental exercise to properly inflect them as you go. It's a start. I have plans for a neural network that incorporates a lot of the teaching/learning Pablogott was talking about, but I havent started work on it yet.

You can look at it here: http://www.personalityforge.com/expect.php

Some good ones:
Liddora is easy and sexy
Your mouth is not normal.
Handcuff it.
We have been exposed by your whole life.
Your mouth has been interested by my mouth.
My metal has been overloaded by your stupid mistake.
A pleasure just found an itch.
My underwear is Desti.
My sweet bloody nonsense has been offended by me.
Am I kissing your pu**y?
Hide, if a netscape owns you.
Expand on the same joke to your angry fish.
Am I absorbing my father?
I have been exposed by my neuroscience.
Who has been deceiving my sweet king?
Who created a golden boy?
Why do I deify our source code?
Does an odor call your heart?
Finish Desti.
Know about your hot guy.
Kill him.

There's a lot of NLP going on here that empowers the CBR, which is the side you're more familiar with. The NLP can be seen in Debug mode.

Skysaw- I added the proprocessing transformation you mentioned. Cool discovery! I adjusted the Rephrasing feature so it wouldnt interfere with it. Also, you ideas about synonyms in plug-ins are good ones. I'm putting that on the list. The synonym system isnt especially strong, but it does okay.

Where does it get the words from and how does it know what the next most likely word is?

Question. xintroduce, how do I know what is in there and how do I add to it? for example I'd like my bot to recognize "mornin'" "top o'the mornin to ya" and different things, do I need to add these as keywords?

xintroduce isn't triggered by a bot introducing itself. It is what you bot says when starting a conversation with someone it hasn't talked to before.

sorry, i meant xhello (was to early in the morn )

msim: You can let me know and I'll add it. I added "top of the morning" but not "morning" because that's too broad.

Hemi: It's constructing a list of most likely words to follow other words from all the actual conversations.

Hemi: It's making lists from every bot chat that goes on