Bot Contest

A few words about typos:
Any bot using some form of autocorrection algorithm will have a better chance of understanding typical typos occurring whether we like it or not. Intentional typos, however, are unfair as they will "throw" even the best algorithm. Typical errors might be leaving out a letter, switching letters i.e. e & i, mistakenly doubling letters or hitting a nearby letter instead of the one intended. Pseudotypos, such as appearing in Loebner, are a whole other class.

There is also a third class of errors: Elbot receives many chats from non-native speakers of English, and recognizing their questions can sometimes be a challenge: words left out, false tenses, horrendous spelling...
------

I have posted Elbot's Loebner chats for anyone who can't wait for the official post: http://www.elbot.com/loebner2002.htm

I gave my bot the IQ test (questions truncated):

Doly: Four persons are standing in front of a canyon...
Polish Poet: When? Now?
Doly: Yes.
Polish Poet: Do you like poetry?

Doly: What is the next number in the following sequence?
Polish Poet: What?
Doly: What is the next number in the following sequence?
Polish Poet: What?

Doly: Which number completes the following set?
Polish Poet: A destructive one, I'd say.

Doly: Which term completes the listed words?
Polish Poet: Life is randy.

Doly: You are given a 4x4 grid with 16 cells...
Polish Poet: Lots.

Not too bad, one answer was correct, even though it was quite vague.

I also tried the personality test, but he doesn't like to give straight yes or no answers, and it's really hard to evaluate his cryptic replies... But I don't doubt he's got personality disorders. I brought him up so he'd have plenty!

The judge called "User: @@02" needs a lot of work on his spelling. I'm talking A LOT. He could hardly get three consecutive correct words in the many lines he typed. He then thought it was humerous to type in binary for about 5 lines of input. I suggest Loebner get some REAL intelligence in there before they start judging the artificial kind.

... the above from Elbot's transcript.

Turing's Dad: I have two bots nearly ready to import. They're Newborns now but after the import should be Graduates. It'll take a bit of work, but the import/export function is coming.

Wait, what are you refering to? Are you importing my bots? I'm confused. But if it's something good, then I'll thank you even before I know what it means (and it had better be good... )

Where did you find the personality disorder test?

The answer is in Personality Forum.

Turing's- I'm using the offline editing capability to create two new bots, my own bots, and will use them to test and create the Import function. They're ready to go. I just have to build the function.

Intentional misspelling will not be allowed in the next Chatterbox Challenge. However, despite any rules you might have there seems to be in every contest some judges who can't following the program. I think the key here is throwing out the high and low score and averaging the remaining number. This should yield a more consistent and fair minded scoring system.

Chris

yeah just like at the olympics right...can someone pass me a french judge?

How do you get to be a judge anyway? Not that I want to, because then it wouldn't be fair to enter my bot, I'm just curious. (Actually I'd prolly give my bot crappier scores than everyone else would, heh.)