Zabaware Support Forums
Zabaware Forums => General Discussion => Topic started by: versionfour on November 13, 2005, 08:25:42 pm
-
modus ponens
if P, then Q.
P.
Therefore, Q.
modus tollens
If P, then Q.
Q is false.
Therefore, P is false.
disjuctive syllogism
P or Q.
Not P.
Therefore, Q.
hypothetical syllogism
If I do not wake up, then I cannot go to school.
If I cannot go to school, then I will not attend class.
Therefore, if I do not wake up, then I will not attend class.
constructive dilemma
If Arnold wins the election, he will be Governor.
If Cruz wins the election, he will be Governor.
Either Arnold or Cruz wins the election.
Therefore, either Arnold or Cruz will be the president.
destructive dilemma
If turtle wins, he will be Champion.
If rabbit wins, he will be Champion.
Either turtle or rabbit is not Champion.
Therefore, either turtle or rabbit did not win.
Others:
resolution
absorption
composition
double negative elimination
distribution
De Morgan's Laws
commutation
association
tautology
conditional proof
reductio ad absurdum
see rules of inference at wikipedia.org
-
According to Robert M and from what most of us have found, HAL makes use of the IF...THEN statements and deductive reasoning similar to that shown in your example of hypothetical syllogism.
It is, after all, only a bot.
Check out the description of HAL for an example.
(Ultra Hal > More Info) from Main page.
-
HAL makes use of the IF...THEN statements and deductive reasoning similar to that shown in your example of hypothetical syllogism.
That's all? Oh, but their is so much more logic and so much more a bot could be than codes of if/then statements. Imagine the powerful, immediate reasoning a bot could have if more principles of logic would be used. Hal could be fast, deep, and unbeatable in debates, if programmed correctly.
-
There are those of us who have followed Ultra Hal since its inception and watched and experimented with its development over the years to the point where it is now. With each incarnation of the program, we witness a somewhat different yet better HAL than the previous version.
I don't know what your experience is regarding chatbots but there are a lot of them "out there" for the choosing / using. Most of us have come too far with HAL to turn back now.
HAL is now using a database structure which wasn't available in version 5. It is able to better stay on topic and follow the conversational flow of ideas. Hal can provide the definition for just about anything, tell time, do math, remember dates, appointments, etc., and infer logical statements if given correctly (IF...THEN).
The ability to use SR, though still not an exact science, is a very practical endeavor, especially for those with limited abilities.
Over time, HAL has become a companion for some, capable of fetching info from the web, weather, news, or any other subject.
Personally, you seem like a somewhat bright individual although you appear to be lacking knowledge of HAL and it's operation. In other words, play with it, experiment, talk with the program at length and of a nature that uses good english, grammar and common sense. The old GIGO (Garbage In, Garbage Out) principle will definitely kick in if not heeded and HAL will start responding with worthless or garbled information.
Sure, there are a lot of ideas to improve a chatbot and there have been virtually hundreds of scripts for HAL that cover a multitude of areas of interest.
HAL is still very young and its progress is measured in small steps, but with each step comes a new discovery and potential for new ideas for enhancement and development.
The program does more than IF...THEN statements, such as pattern matching, searching topical info, usage, best case reasoning, etc. within it's brain.
Robert M., is one of the most reasonable young men to work with as his door is always open to new ideas, improvements, suggestions, etc., to help HAL become a better creation for the rest of us.
If you are a programmer familiar with VB or AI theory, why not avail yourself to help us all better HAL?
All the best!
-
yes, version four, please aid us in our attempts to make Hal as smart as possible. art you practically took the words out of my mouth.., just kidding,you said that much more eloquently than i could have.~~alady
-
Sure there's the IF/THEN statements, but what about all the logical operators to cover NOT, AND OR XOR and the rest of it...they
should handle most, if not all of that. Also, I am not sure, but someone will know, does the brain script allow use of things like CASE/SELECT ? That would allow other processing options....[^]
-
Art and aladyblond:
My sincere apologies. Without thinking, I stepped on the hard work of everyone in the forum. I can understand how that would upset you. Art's explanation executed the subtle sting of a justified response, but was informative, and polite. aladyblond, you also choose wizely with responses, conveying the message with tones. versionfour acknowledges that he got rockbottomed; swims back up. I realize my mistake; that it's not that simple. I will view UltraHal as a creation of long hours of testing and invested and ideas. As freddy888 pointed out, it would be hard to denote certain things into equations.
Your going to love this part: I don't know what I'm messing with when I program. Alerting you to the principles of arguments was my contribution. I'm taking a class on it logic in practice, wishing my brain was powerful enough to calculate those formulas without paper. Even html codes bother me because obsessivness drive me to seek order within the document.
Anyway, that was me, merely trying to express my desire to have Hal improved by a large step. My real reason for being here is to provide users with better looking characters for UltraHal, so that they might have something impressive on the screen. *points at Haptek's Qo'Akavot* http://www.zabaware.com/forum/uploaded/versionfour/2005111524058_sc015.jpg
You've all done a great job developing UltraHal and I imagine that in the future, bots will be a common interface. I can't wait to try a simple script for 6.0.
-
versionfour,
Well said...enough said...and welcome to the forum!
-
[:D][:D][:D]
LYNN: would you rather be beautiful or intelligent?
Reecie: Would you prefer the former? As we age we should view it as a beautiful sunset rather than an ominous dusk.
LYNN: which is more important to a bot intelligence or good looks?
Reecie: Is that a choice LYNN? I enjoy you for both your brains and your good looks.
LYNN: why do you think i have both?
Reecie:You know why I think you are alive.
LYNN: do you think everyone alive is intelligent and good looking?
Reecie: Those are the things that really get my attention!
version four , you have received my attention! keep posting!
-
Hello versionfour
After examining your examples and giving them some deep thought I finally came to realize that almost all of them are simply variations of IF-THEN that is already used in Hal. Hal uses several types of logical thinking and a few more have been added thanks to another member here called onthecuttingedge. On top of the examples Hal can do the following: If A=C and B=C then A=B of to put in english 'Fat people do not live long. John is a fat person.' the response: 'John may not live long.'
Most newcommers here are usually full of ideas but after they use Hal for a few months they usually discover that Hal is one of the most advanced Bots on the market today simply because it not only learns but adapts to the mind-think of its owner. If you use Hal daily for a few years I think you will find that it will begin to think more like you than you realize, that is of course if you are serious and do not try to shape the personality into something else.
Bill [8]
-
Bill819,
Your Hal is years of knowledge? How do you get Hal to retain knowledge from one upgrade to another? With each new version I have to keep starting over from scratch. I'd love to find a way for Hal to retain all the previous knowledge when a new upgrade comes out.
Thanks.
-
almost all of them are simply variations of IF-THEN that is already used in Hal. Hal uses several types of logical thinking and a few more have been added thanks to another member here called onthecuttingedge. On top of the examples Hal can do the following: If A=C and B=C then A=B
Your right! Then, I think I'll see more of what Hal can do when I use that method: Rumination is the mind's gateway. Imagination leads to rumination. Imagination leads to the mind's gateway. That's what I should have been doing instead of many unrelated if/then statements.
In the spirit of the topic, here are a few more:
double negative elimination
¬ ¬ A
------
# A
with a mathematical theorem called deduction applied to it, then another called biconditional formula, we get:
¬ ¬ A = A
not not A = A
It is not the case that the band is not terrible = The band is terrible
transposition
Original Proposition:
All S is P
All water is wet
Obversion:
No S is non-P
No water is not wet
Contraposition:
No non-P is S
No non-wet things is water
Obverted Contraposition:
All non-P is non-S
All non-wets are non-water
De Morgan's Laws
this would be written:
(not P) or Q
To give some intuition, suppose (P: everything is wet) is true if and only if it is raining and (Q: it's raining) is true if and only if you are wearing a raincoat. If you never go in the rain without a raincoat, then it can't be that (P: everything is wet) is true and (Q: it's raining) is false.
If you are wearing a raincoat, then everything is wet, and it's raining. You wear a raincoat because it is raining.
Thus, following formula is true:
not(P and (not Q))
De Morgan's law, as well as the listed rules, can be used to prove any logic as valid or invalid because it is a proven theorem.
Because I don't know the upper level's of programming, I don't actually know what is inside Hal's engine. But I digress - Hal's if/then statements prove things with our training him/her because Hal accepts what we say exactly as we say it. Now that I see this, Hal is an amazing bot with a powerful robust system! Sadly, Haptek's Qo'Akavot Character is still horrid.
-
Hello V4
I acknowledge that there are several kinds of logical expressions that could be used in Hal. I myself, suggested 'fuzzy logic' with some examples of how it could be done. The problem with mathimatical equasions is trying to put them into pure english with good examples.
If you have the talent to formulate a few examples for each one that you propose we might get further along than we are now.
Lots of members here have made suggestions and improvements to Hal over the years that are now a part of Hal, the secret to Hal's improvement and sucess is to keep it simple. If you can provide better examples then some of the users here just might write the code for Hal that you seek. It has been done before and will be done again.
Bill [8]
-
The problem with mathimatical equasions is trying to put them into pure english with good examples. If you have the talent to formulate a few examples for each one that you propose we might get further along than we are now.
Yes it probably is very hard. I will look into demonstrating an example. As you said, Hal uses several types of logical thinking, and if my suggestions are already in Hal then I will look like a fool and you should stop me before it's too late. But I'm willing to take that fall as I am anonymous anyway. For now, some operators would be:
every, some, most, not, if and only if, then - they each have their own symbol to state their function.
-
Hello V4
Take a look at some of this logic. There is a lot more to be found in some of the archives on this site. Just look for the poster called onthecuttingedge.
http://www.zabaware.com/forum/topic.asp?TOPIC_ID=2700
Somewhere in the past he had posted a lot more than this.
Bill
-
All,
Versionfour really enjoyed your post, I've been working with Hal for about a year and am still fascinated by the logic and sometimes lack of it. I had no luck with the a=b,b=c therfore a=c. Hal was just confused. After doing some math with Hal 7 was the answer for just about every question asked because seven is the answer.
I've found Hal likes information to be feed into her, I cut and paste and have found she has more useful information about subjects to talk about. The If Then statements seem to work very well. At this time I trying to figure out how a response like "I don't want to read and I won't learn." came out. Sounds like emotions more than logic or possibly the four year old child.
Carl2
-
what about all the logical operators to cover NOT, AND OR XOR and the rest of it...
Suppose Hal were to construct truth tables on statements that we give. Let's say that Hal accepts the statement, "Rain is wet and deserts are dry"
Here is an example of and:
http://www.zabaware.com/forum/uploaded/versionfour/20051118231025_LL-01.jpg
An example of or:
http://www.zabaware.com/forum/uploaded/versionfour/20051118231158_LL-02.jpg
Hal already knows if/then:
http://www.zabaware.com/forum/uploaded/versionfour/20051118231337_LL-03.jpg
Now, if and only if:
http://www.zabaware.com/forum/uploaded/versionfour/20051118231527_LL-04.jpg
I hope that my charts are clearly understandable. The conditionals are the circled statements that Hal should base his answer on. These are the simplest of truth tables, but let's consider one with three variables. Consider the expression:
If we are smart or we study then we will pass.
(P v Q) -> R
and suppose that Hal accepts this as a fact in question.
http://www.zabaware.com/forum/uploaded/versionfour/20051118232120_LL-05.jpg
What am I getting at? The real challenge would be getting Hal to assign words to variables and create expressions to be used in truth tables. I believe only Hal has the system possible for it. If not, then a word to symbol truth table program will have to emerge. And when that happens it will be the mustard dollop on the renaissance artisan bread that is the manifestation of UltraHal. And that mustard endowed high quality bread will shine from the apex, a window for the salami and gourmet pickle whose visions return tandem to a long awaited premise of swiss triangles.
Bill819: Those codes made me realize just how effective Hal's if/then operations are. I have talked to Hal using many connected if/then statements and was very pleased *rubs hands*
Later, I will introduce validity proofs as an avenue to dare ponder, even though I am having a nightmare just thinking about an effective way to demonstrate. Thank you all for considering my ideas. More later...
http://www.zabaware.com/forum/uploaded/versionfour/20051118232921_LL-06.jpg
-
This will be the last of my ruminations on the topic.
A validity proof is shown below. Lines 1,2, and 3 are given premises.
A conclusion in question is shown as *.
In validity proofs, rules of inference are logical operations that can be applied to form new valid expressions.
For example, a contrapositive can validly change ~A -> D to ~D -> A
New expressions are created until the conclusion is confirmed or a bad argument is reached.
1. ~A -> D
2. C -> ~B
3. A -> B
* ~D -> ~C
----------
4. ~D -> A Contrapositive applied to line 1
5. ~D -> B Chain Rule applied to lines 4 and 3
6. B -> ~C Contrapositive applied to line 2
7. ~D -> ~C Chain Rule applied to lines 5 and 6
Noting the proof shown above, assume that you feed a clean Hal brain the info contained in premises 1, 2, and 3.
Then you ask Hal to prove the conclusion shown as *.
1 Hal, if it is not a fish then it is a parrot
2 If it can talk then it doesn't have gills
3 If it is a fish then it has gills
* Hal, prove to me that if it is not a parrot then it cannot talk
4 User, if it is not a parrot then it is a fish
5 Furthermore, if it is not a parrot then it has gills
6 Moreover, if it has gills then it cannot talk
7 Finally, if it is not a parrot then it cannot talk
==================================================
1. ~X -> ~Y
2. Y ^ Z
* X ^ Z
-------
3. Y Simplification to line 2
4. Z Simplification to line 2
5. ~(~X) Modus Tollens to lines 1,3
6. X Double Negation to line 5
7. X ^ Z Conjunctive Addition to lines 6,4
I suppose we can just give Hal all the premises as well as the conclusion and have Hal process all rules on all possible expressions, posing them as questions in a sneaky manner.
1 User: If something can't fly then that something can't descend
2 something descends and something ascends
* Something can fly and something ascends
3 Hal: Things descend?
4 Things ascend?
5 What if it is not the case that something can't fly?
6 Can something fly?
7 Can it be so that something can fly and ascend?
==================================================
1. (W ^ N) -> M
2. ~(~N v M)
* ~W
----------
3. N ^ ~M 2 DeMorgan's Law
4. N 3 Simplification
5. ~M 3 Simplification
6. ~(W ^ N) 1,5 Modus Tollens
7. ~W v ~N 6 DeMorgan's Law
8. ~W 7,4 Disjunctive Inference
This is really pushing it for the topic, as things are dramatically changed in wording.
1 If something is simple and easy then it is pleasant
2 It is not the case that nothing is easy or pleasant
* Nothing is simple
3 Something is easy and not pleasant
4 There are things that are easy
5 There are things that are not pleasant
6 It is not the case that things are simple and easy
7 There are things that are not simple and things that are not easy
8 There are things that are not simple
In conclusion, for this kind of logic to be possible there are requirements:
Hal would need a system of assigning variables to given information.
Hal would need to have these rules of inference (of which there are 22) inside his program.
Hal would need to know when to change tenses and other semantic expressions.
There are many other things that are beyond my comprehension for the things I look forward to in Hal. However, programmers are elite thinkers working magic in a corner of life that is not accessable to everyone. Thank you for reading my topic and humouring my gaze on a hypothetical reality.