[MUD-Dev] Chatbot

[rayzam]

----- Original Message -----
From: "Aaron Mulder" <ammulder at alumni.princeton.edu>
To: <mud-dev at kanga.nu>
Sent: Thursday, July 05, 2001 4:43 PM
Subject: Re: [MUD-Dev] Chatbot

> On Thu, 5 Jul 2001, Michael Tresca wrote:

>> Thought you might find this interesting:

>>   http://aimovie.warnerbros.com/

>> On the A.I. web site, Chatbot (ALICE, click Enter to see it)
>> gives clues about the movie.  I think.  Anyway, the Chatbot is
>> sufficiently advanced that several people on the discussion
>> boards thought they were talking to a real person, or that "a
>> real person was answering for the 'bot some of the time."

>> I'd love to make our NPCs that believable.  I suspect the
>> audience has something to do with it (e.g., mundanes not used to
>> chatting to programs).

> This is exactly where I was going with the language parsing
> thread.  Step one is for the NPC to understand what you're saying
> to it.  Step to is for it to think of something coherent to say
> back.  We'll just take this one step at a time.  :)

> BTW, the parsing of sentences is coming along well.  It takes a
> surprisingly small amount of code, and then a big dictionary.
> Still have a ways to go on attaching the concepts to the words.
> I'm trying to figure out how to modularize the words and concepts
> so they don't all have to be loaded at once, and somehow find a
> good balance between all NPCs sharing the same data set (which
> performs well but doesn't distinguish them much), and NPCs all
> having a unique data set (which makes them very personalized, but
> required a ton of background for each one and isn't very
> efficient).  

Aaron, while I suspect you may know of this project, I expect others
may not:

  http://www.cogsci.princeton.edu/~wn/

WordNet:

  "WordNet is an on-line lexical reference system whose design is
  inspired by current psycholinguistic theories of human lexical
  memory. English nouns, verbs, adjectives and adverbs are organized
  into synonym sets, each representing one underlying lexical
  concept. Different relations link the synonym sets. "

I've seen it in action, and it's quite fast. A relational system may
be the way to handle concepts and modularizing...

    rayzam


_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev