While I am brand new to this forum, I have been reading some of the threads for awhile and always find them very interesting.

I started on a “simple” “easy” no brainer chatbot about 6 years ago as a game for my daughters.  The project sort of grew, and grew and grew lol.  It has been and continues to be a wonderful learning process.

The current version of “Nicole” I would very much like to enter into this years Loebner contest.  However I am having trouble understanding what I need to do for the Loebner Protocol.  ANYONE willing to give me any information will be most welcome!!

This last year has been a difficult one for our family, my wife was diagnosed with cancer.  So I came up with the idea that the family could focus on getting Nicole ready for the Loebner, to allow us not to be so deeply steeped in her cancer.  The prognosis for my wife is excellent, the chemo and radiation has taken a heavy toll but she alive and doing great.

So this simple “no brainer” program has been anything but over the years.  As a “family” project, it has brought us closer together and caused some for lively discussions.

Nicole (my bot) did win the best new bot program in the latest chatterbot contest.  However to be honest that was more luck than the programs abilities.  I just got lucky on the direction of the questions.

Thank you for allowing me to be a member of this community!
Jim

You’re welcome. Please join any of the LPP topics in the Loebner Prize subforum, because I’m sure you’re not the only newcomer looking for these answers. Documentation on the Loebner Prize Protocol is rather disorganised, but here is all there is on it:

http://www.paulmckevitt.com/loebner2013/loebner2013call.txt
http://www.paulmckevitt.com/loebner2013/loebner2013entrantnote.txt
http://www.paulmckevitt.com/loebner2013/loebner2013judge_program.txt (rename extention to .pl and run with a perl interpreter installed on your computer)

What your chatbot has to do is create numbered subdirectories for each letter of output. The LPP judge program will scan for these and take them as input, so receiving your chatbot’s message letter by letter. Vica versa, the LPP judge program creates numbered subdirectories for every letter that it sends to your program, and your program must take those as input. These subdirectories are created in a specific directory, referred to as the “communications folder”.

Holy crap, this is a really terrible protocol.

Don thank you for your reply, it helps.  Jarrod, I totally agree with you and believe time could be better spend on improving our bots.  However, it is what it is.  One can dream that someday the protocol will be much easier and better.  But I put that into the same category as my bot carrying on an intelligent conversation for 20 minutes, highly unlikely.

Through extensive testing I have determined my bot only has 3 problems:

1.  The questions it can’t answer.
2.  Program bugs introduced by yours truly
3. The inability of the program to understand what humans are saying.

Aside from the above my bot doesn’t do bad. 

Jim

I believe those problems are present in just about every chatbot I’ve ever seen, so , no reason not to test it.