Archived

This topic is now archived and is closed to further replies.

Diego_Hawthorne

I have a few questions on a conversation bot...

Recommended Posts

I am starting a conversation bot. Now, before anyone asks, this is just to learn more uses of, and to gain experience with, AI. I am not in school, nor have I ever taken a course in programming. My question is not even a question, but a course of action in a way. I was wondering, without dropping specifics, what hurdles should I be prepared for? I know that a decent database will be required. However, I would be using the player''s speech as the computer''s database. I am going to use a few basic filler-type words, as well as questions and a boolean organiser for targets(you, me, I, they, we, etc.). But, as I''ve said, this will let the bot create it''s own database of likes, dislikes, fears, joys, etc. My problems are many(as I''ve never done a project like this before)...but I would like guidance and tips, instead of code, or outright ''here''s how to do it'' responses. Thanks in advance, as this site, gametutorials.com and gamasutra keep me going Hasta!

Share this post


Link to post
Share on other sites
Search google for "alice", "win alice" and "chatter bot" and you should find a wealth of information pertaining to conversation bots and maybe if you''re lucky a few references regarding programming hurdles and such.

Share this post


Link to post
Share on other sites
quote:
Original post by bishop_pass
SHRDLU kicks ass over any ''conversation bot''. SHRDLU was early 1970''s technology. But SHRDLU had discourse and domain knowledge. Do conversation bots?




Very SIMPLE domain knowledge, mind you. There are certainly far better discourse engines out there these days.

Timkin

Share this post


Link to post
Share on other sites
quote:
Original post by Timkin
Very SIMPLE domain knowledge, mind you. There are certainly far better discourse engines out there these days.

Timkin

It is far more impressive than anything ever demonstrated to most on these boards. Perhaps you would like to share some links? The point is that SHRDLU models discourse and makes use of domain knowledge. Your comment sounds as if you are belittling something designed out of ''70s technology.

Share this post


Link to post
Share on other sites
quote:
Original post by bishop_pass
Your comment sounds as if you are belittling something designed out of ''70s technology.



My apoligies, it wasn''t mean to sound that way. Perhaps by not elaborating or justifying my statements I left them open to misinterpretation.

For what SHRDLU did, it was very impressive, particularly given the technology it was implemented on!

In terms of current systems though, the Blocks World domain is not what one could consider complex. There has certainly been more recent (than 1971) success in more complex domains, particularly in natural language generation from spoken discourse and in NLG for text document generation.

As for resources on this material, a quick search of citeseer will provide hours of reading.

Regards,

Timkin

Share this post


Link to post
Share on other sites