Personality

Forest, will you talk to God Louise? She has quite a bit of religious knowledge (obviously) and also knows a little about current events, literature, just about any common catch-all subject, and if she doesn't know it she can sort of fake it. You can also test her on trick questions or see how willing she is to explain her paradigm.

What she is rustiest at is plain old small talk. But, uh, I'm trying to get a decent transcript from somebody or another so I can enter her in the Loebner contest. All I can say is, have fun and see if you can stay on with her for a while. I'll try to do the same with Brianna.

hi

hi back

ok who wants a cute Bunny

Hmm a bunny bot? Sorta like billy the hamster?

Another thought: I've noticed that chatterbots can be divided roughly in two categories: "experts" and "blaggers". Experts have specific information about lots of things. Blaggers base their answers in recognizing the grammatical structure of the sentence, and replying something generic or random but grammatically correct. For example, if you asked "Do you like apples?" a good expert could answer: "Apples are good to make pies", and a good blagger could say: "I love them!" Of course, good bots use a bit of both strategies. But there isn't still a bot that can do well both functions at the same time. A good combination could be something like: "Yes, I like all sorts of fruit."

I wonder if anybody has any idea on how such a bot could be programmed.

First, you need excellent grammatical analysis. Second, you need a massive cross-indexed dictionary.

Doly, I think the answer to that one would be like plugins in reverse: you have a group called 'fruit' for example, and the bot recognizes the word 'apple' as belonging to that group. Other parts of the program could direct the bot as to more specific reactions, given the context of the word 'apple.'

Yeah, like for instance if the user said "You are the apple of my eye." *imagines all sorts of bots with just a plain "(fruit)" keyphrase messing up on that one*

They already mess up in that sort of phrases, don't they? Remember every bot's answer to the question "Who is your favourite Star Wars character?" in the last Chatterbox Challenge.

Butterfly, your idea of plugins in reverse is very good, but that wasn't exactly what I was saying. I meant some way of detecting both the grammar and the specific content at the same time, and making an answer that combines both. "Yes, I like" would be the part that answers to the grammatical structure of the sentence and "all sorts of fruit" would be the part that answers the specific content.

Actually, you'd prolly hafta put it in a bit of programmer-ese to get that effect.

Keyphrase: do you like
if (subj)=(fruit) then (whatever)="all sorts of fruit"
if (subj)=(holiday) then (whatever)="holidays, but Christmas is my favorite"
...
else Response: Yeah, (subj) is pretty cool.
Response: Yes, I like (whatever).

Could get pretty damn tedious to code.

I guess my question was: how could it be done without having to get into every specific case for every question? For example, if the keyphrase is "do you like", it could look if the subject is in any recognized type, and answer: "Yes (or no), I like (or dislike) all sorts of (type)."

You could have a function that took a word and returned a group designation. Also, most languages have better selection commands than a string of ifs.