Personality

We really need a way to escape certain words and phrases from spellcheck and preprocessing. Should be easy enough to implement. In addition to the examples given above, my attempt at using 'Konnichiwa' (Japanese greeting) as a keyphrase died in spellcheck also. This kills a lot of useful and interesting keyphrases.

*looks at Shadyman*

Are you sure about that? I think you might be trying to trick me!

Quick question on seeks...

I have a keyphrase for "What is your cat's name", to which Duo replies with: "His name is so and so, didn't I tell you about him?"

I have a seek for "no" - to which he'd follow up with more information. However, the seek keeps getting ignored.

Like so:

Guest: What's your cat's name?
Duo Maxwell: His name is Jupiter. Didn't I tell you about him before?
Guest: No, what does he look like?
Duo Maxwell: What are you talking about?

Would a seek for "look like" fix this problem?

*nod*
or a seek for "what does" or a seek for "look", you get the idea

Yay - it worked. Thank you!

(subj) vs. (postkey)

Which one would you consider better? Or, is there really a difference?

In the past, I always used (subj) to reply to keyphrases such as: "You are"

Response: What makes you think I am (subj)?

So, if someone said "You're funny," Duo would reply with "What makes you think I'm funny?"

Recently, though, I've only been using (postkey). Should I go back and get rid of all the (subj)'s? (They seem to be working perfectly fine, so far.)

I think (subj) is eventually going to get phased out, so you might as well change them. (Yeah, it'll be a big job for me, too.)

subj isn't necessarily what's after the keyphrase though. it's safer to use postkey (or a (*))

Will (definition) work for (*) or only for (subj)? I see in the Book of AI that the (definition) help references (subj).

Uh, I think you put it in a keyphrase like

What does (*) mean
response: (definition)

Slowly changing all the (subj) to (postkey). It's going to take a while.

Oh. Duo seems to be depressed/sad and from reading his transcripts, I can't figure out why. All the dialogue has been rather pleasant - yet he has a next to his mood.

What affects a bot's mood if not the conversations they have with people?

/curious/

DaMoyre, thanks for posting that problem above- "no" should have worked just fine but I apparently hadnt hooked up the new "yes/no detection function" up to the Seek search. It's now working as it should.

Also, (subj) isnt equivalent to (postkey). It means (key1) if there's a "*" in the Keyphrase and (postkey) if there's not. The old way didnt distinguish between them- you only had access to (subj), which is why it's not ideal any more.

The Mood is a fickle affair. It's a mix of the bot's overall Comfort, Popularity, and the Mood from the last conversation he's had. If he is generally happy but just talked to someone he hates or who was mean to him, his Mood will drop, but come back up later.