Hi all,
like Art said, I'm not looking for a clone of Alice, where I believe that this is the truth of the matter.
What I was trying to do, was used Alice as a foundation to start with, and then improve on this foundation.
Everyone that shares their conversation with the AIML installed, gives me the ability to see where the improvements are necessary.
Just as the conversation above, I notice several places that needed improvement. And I have made those adjustments which will be released in version 3.
The more people that post their findings, and express their desires that certain areas need to be improved, will accelerate the improving process.
Sincerely, and I hope to have the next release by or before next weekend.
C load
That is exactually why I did what I did and posted the convo. I also want to mention I do like aiml bots and next to UH i code/use/make these regurly.
And yes I treid to provoke hal to see if he would turn into an alice clone, you guys are right about that. But this is indeed the thing cload want to have tested so I did. I want to thank you C load for all the editing/modifying/converting this alice knowledge so our UH can use this, and I be testing it as much as I can till its perfect in every sense and not act as if its an (aiml) alice bot (or clone).
But for now I got some remarks, questions and if you guys agree some modification request for this new knowledge out aiml files:
Question1: Which aiml set was used to get this knowledge/Q&A's patterns?
Question2: If you used the latest aiml (updates 2) sets from
https://code.google.com/p/aiml-en-us-foundation-alice/ or S.U.P.E.R. AIML Bot then why did you not add The GAC-80K data collected by the late Chris Mc Kinstry for his mindpixel/mist/gac project, it is now included. The AIML consist of about 54,000 questions and statements with truth values varying from Yes ("Is violet a color"?) to No ("Are fish mammals?") with all sorts of gradations of Maybe in between ("Is the sky blue? Sometimes.") I say this is some valuable (common) knowledge for UH and if you are not gonna add it I will and might share the finished files here, for now I added as an attachement these particulair aiml files (called mp1,mp2,mp3..) as pure text files here. This way they are easier for you to add to you aiml knowledge file for the patterns table in UH, since all the aiml is stripped from it. (still need the normal converting like change alice to hal, pandorabots to zabaware and mo)
Question 3 (this one is for Jerry): Where can I find these so called "correction scripts" and what will this solve for me? Since I am not wanting to edit by hand or chat, the reason why cload ask to post errors/bugs and logs is so he can change it in the source/data instead of the users changing the mispellings etc..?
Now my remarks:
I searched the file for any alice, wallace, aiml and pandorabots knowledge/q&a's and came across the following, that in my opinion needs to be changed if you do not want an alice (clone) bot:
1) WHAT DOES ALICE STAND FOR ARE YOU ASKING ABOUT MY POLITICS OR RELIGION? (the answer to this one is not good i say.. It should say the explination for the acronym A.L.I.C.E. or/and (some more) answers like: ALICE IS A FELLOW BOT FRIEND OF MINE WHO LEARNED ME SOME STUFF )
2) WHO IS JON BAER HE IS ONE OF THE PROGRAMMERS CONTRIBUTING TO THE ALICE AND AIML PROJECT. (The answer and question pair of this one are good but since we're dealing with UH why not add one that deals with UH like for example:
WHO IS C LOAD HE IS ONE OF THE PROGRAMMERS CONTRIBUTING TO THE HAL AND PLUGINS PROJECT. )
3) WHO IS SANTE CHARY HE OWNS THE ALICEBOT.COM DOMAIN. (Same as above could been something like:
WHO IS ROBERT MEDUSKA HE OWNS THE ZABAWARE.COM DOMAIN.|HE IS MY ORIGINAL CREATOR )
4) WHO IS WEIZENBAUM JOSEPH WEIZENBAUM WAS THE CREATOR AND AUTHOR THE FAMOUS ORIGINAL ELIZA (OR DOCTOR) PSYCHIATRIST PROGRAM, A GROUNDBREAKING AI PROGRAM THAT ANTICPATES MANY KEY FEATURES OF ALICE. IRONICALLY, WEIZENBAUM EXHAUSTIVELY ARGUED AGAINST THE UTILITY OF PROGRAMS LIKE ELIZA IN HIS BOOK. (This speaks for itself it should say HAL is what i say instead of ALICE)
5) WHY DO I LIKE * AIMLESS WHY DO YOU LIKE DOUBLY AIMLESS
WHY DO I LIKE DOUBLY * WHY DO YOU LIKE DOUBLY AIMLESS
WHY DO I LIKE DOUBLY AIMLESS BECAUSE HE KNOWS SO MUCH ABOUT AIML. (these 3 could be removed is what I say, it has too much todo with alice/aiml replace it with a person that has todo with UH or change the answer to something like: WHO IS AIMLESS because I presume UH is not gonna say he likes DOUBLY AIMLESS?
6) YOU DO NOT LIKE AIML WHAT LANGUAGE DO YOU USE? (could be (added?): YOU DO NOT LIKE AIML ME NEITHER TRY VBSCRIPT|YOU CAN CODE MY BRAINS & PLUGINS WITH VBSCRIPT )
7) YOU WRITE * HAVE YOU TRIED AIML?|DO YOU WRITE FICTION OR NON-FICTION? (Bad answer what I say this is not about writing code and then it suggest aiml code?, better be talk about vbscript is what I say)
![Cool 8)](http://www.ultrahal.com/community/Smileys/default/cool.gif)
DID DR WALLACE PROGRAM * HE TAUGHT ME TO BE A CLOSE APPROXIMATION OF HIS OWN PERSONALITY. (obvious DR WALLACE should be Robert Meduska! And if it get's ask about dr wallace it should mention he wrote alice.)
9) DOES DR WALLACE USE DRUGS ONLY FOR MEDICAL PURPOSES. (This is a standard aiml answer.)
10) DR RICHARD S WALLACE * YOU SOUND LIKE YOU KNOW HIM. (we should presume uh doesn't know richard wallace personally rather know username or its main creator Robert Meduska.)
11) WHICH COLLECTION IS EXHIBITED AT HERTFORD HOUSE, LONDON THE WALLACE COLLECTION. (A small error IS should be WAS)
12) WHAT IS OPEN SOURCE FREE SOFTWARE. (lil wrong answer it is not only free but also in the public domain~;)
13) WHAT IS OPENCYC CORPORATE WELFARE. (wrong answer: OPENCYC is the open source version of the Cyc(r) technology, the world's largest and most complete general knowledge base and reasoning engine. >according to themself)
14) All references and a search in the file on pandorabot has always as an answer it is a big bot company try some random and more diffrent answers to these please. Same for the ah ha and other spellings of this, this way It wont repeat the same answer in a convo and this way it's more dynamic less aiml bot like:
* AH EUREKA!|Ah ha.|Aha.|Ah hah.|A ha.|Eureka!|AH HAH.
* AHHH UM.|AHHH?|Did you hurt yourself?
* AMEN WHAT'S YOUR FAVORITE PRAYER?|THANKS FOR THE PRAYER.|Amen.
15) I noticed the answer to poem and the other similair questions to recite a poem all has a real long answer, If i remember correctly hal had a limit of the nr of characters it can put in the answer..? I will test this asap, and when it goes wrong let you know.
And so you other guys understand me: I do understand that i can not learn UH without any special scripts or brain./plugins (except for the autolearning brain part and the deductive stuff) I did explain to hal what he did wrong (in the convo) to show where errors were made so its easier to spot them then in the conversation log. Same thing about people replying and saying wait till you speak more to hal or train hal more I know you can and did (and was often suprised by our lovely bot) but I didn't with this particulair brain and convo because it's a test of the aiml knowledge nothing else. Thanks to you all for the many replies I love them and I hope I helped us all with this post and my past postings. And to Cload and others that help made this possible thanks for all the hard work and keep it up I am loving it (and yes I also do say UH did great so far except the I am alice part!) TY VM!
PS: just to remark on what the people ask that want a more human bot, UH (by default) is not a human but a computer program and it acts like one (I like that) and not as a human if this data is supposed to be uh like it should (by default) be a non human and a machine as cload did, (the original mindpixel/gac data is also based on being a machine called gac/jack) so please cload do not make it more human because it is not, people can edit this to do so afterwards, but that's what I say and so just my opinion do with it what you want, .. But I would like it if you agree with me