Ai Forums Home Welcome Guest    Thursday, May 24, 2018
Ai Site > Ai Forums > Language Mind and Consciousness > Chatterbot Eliza 3.0 Last PostsLoginRegisterWhy Register
Topic: Chatterbot Eliza 3.0

posted 3/18/2009  15:44Send e-mail to userReply with quote
I'm on the process of writing a new chatterbot in C++ once again, this new chatbot is meant to be a greatly improve version of my previous chatterbot (chatterbot eliza 2.0 : that i have published over the web at:,, etc. One of the goals on these new chatterbot is to have a better encapsulation on the functionalities of the program by using more class (ex: we could use a class for the chatbot responses (class CResponse() {}), one for the inputs of the use (class CInput() {}). I also wanted to extend the database of the chatbot to make it more bigger maybe 5000 lines or more (right now it is: 1376 lines). One of the other goals is to use 'Breadth first Search' for the keyword lookup inside the database. And finaly one of the major goal of the these new chatbot is to have a version the program that is much more faithful to the original Eliza but also have some new functionalities.

 Intelligent Chatterbot Blog
Last edited by Leibnitz @ 3/18/2009 3:48:00 PM

posted 7/12/2009  01:40Send e-mail to userReply with quote
You write in your blog:

When the programs detect new keywords that are not part of the "script file", it saves it on the file "unknown.txt"

May I suggest saving the questions that triggered that event, in toto - and use them on the NEXT interviewee, and then start collecting the *human* answers to it?

Could just simply play it by numbers, the answers given most often could be considered 'best'.

Make the database a fluid affair, not static, and essentially pass the answers through a kind of Darwinistic filter - provide by the *other* people it converses with.

Hope that makes sense. :)

posted 7/12/2009  02:38Send e-mail to userReply with quote
Another option would be to use a Google:

Q. Tell me, why would they run away? (first question, out of the blue).

A. Not so loyal, I guess.

The first return, from Yahoo Answers could give you enough material to create that 'answer' - without a database, even. Just look for 'declarative statements' immediately following that question, elsewhere.

We all borrow snappy comebacks, so why would an AI not do likewise?

posted 10/3/2009  22:51Send e-mail to userReply with quote
Hi. I am also creating chatbot but i am using Delphi and MySQL for this. I dont know, maybe we can work togeather on some A.I. project? :D

 Temp homepage
'Send Send email to user    Reply with quote Reply with quote    Edit message Edit message

Forums Home    The Artificial Intelligence Forum    Hal and other child machines    Alan and other chatbots  
Contact Us Terms of Use