Formalizing Natural Language Semantics

What did you say? And what did you mean by it?

Moderators: AMod, iMod

Skepdick
Posts: 14366
Joined: Fri Jun 14, 2019 11:16 am

Re: Formalizing Natural Language Semantics

Post by Skepdick »

PeteOlcott wrote: Wed Mar 04, 2020 6:02 pm We define this as the body of analytic knowledge: (overcoming Quine's objections)
Analytic knowledge: The set of all knowledge that can be completely expressed using
language and verified as totally true entirely based on its meaning thus not requiring
any sense data from the sense organs.
You are trying to re-invent denotational semantics: https://en.wikipedia.org/wiki/Denotational_semantics

The very thing that doesn't work in practice, except grammatically.

Quine's objections are valid. Analyticity is circular.
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Skepdick wrote: Wed Mar 04, 2020 6:31 pm
PeteOlcott wrote: Wed Mar 04, 2020 6:02 pm We define this as the body of analytic knowledge: (overcoming Quine's objections)
Analytic knowledge: The set of all knowledge that can be completely expressed using
language and verified as totally true entirely based on its meaning thus not requiring
any sense data from the sense organs.
You are trying to re-invent denotational semantics: https://en.wikipedia.org/wiki/Denotational_semantics

The very thing that doesn't work in practice, except grammatically.

Quine's objections are valid. Analyticity is circular.
The set of all knowledge that can be fully expressed entirely on the basis of the meaning
of its words excludes any aspects of sense data from the sense organs. is not circular at all
When you try to find a counter-example you fail because none exist.
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Scott Mayers wrote: Wed Mar 04, 2020 6:00 pm
PeteOlcott wrote: Tue Mar 03, 2020 10:39 pm
Scott Mayers wrote: Tue Mar 03, 2020 9:01 pm
We already recognize that you can solve any particular problem by careful exhaustion. What can't be done is to have a finite machine be able to FIND ALL the possible programs as 'haltable' finitely.
What I have accomplished is deriving the complete encoding (written in x86 machine language) of a halt decider that decides halting for the set of halting problem counter-examples as defined by the Peter Linz H_Hat.
My background is in machine languages. But I'm confused at what you assert such a program is doing up front. Turing presented a closed argument for ALL finite machines, which includes any machine and its languages built upon them. Prior to expecting me to look at such, I'd need to understand more about how you expect to find some/any alternative "halting decider" program that proves other "halting decider" programs used to discover whether you can list all programs that 'halt' has flaws without making those original ones work after all?

I think you might be misunderstanding the original arguments of incompleteness and the degree of their applications.

Another way of describing the problem is,

"Can you find an ideal single universal calculator that can solve all problems with complete satisfaction to solve all possible particular calculator-related problems there exists?"

Are you saying that you CAN, do not know, or cannot (but may have a disagreement about HOW the prior proofs have presented their case)?
I encoded the Peter Linz H_Hat in x86 machine language such that H_Hat correctly
decides halting on itself, something that Linz claims to be impossible.
Scott Mayers
Posts: 2446
Joined: Wed Jul 08, 2015 1:53 am

Re: Formalizing Natural Language Semantics

Post by Scott Mayers »

PeteOlcott wrote: Wed Mar 04, 2020 8:08 pm
Scott Mayers wrote: Wed Mar 04, 2020 6:00 pm
PeteOlcott wrote: Tue Mar 03, 2020 10:39 pm

What I have accomplished is deriving the complete encoding (written in x86 machine language) of a halt decider that decides halting for the set of halting problem counter-examples as defined by the Peter Linz H_Hat.
My background is in machine languages. But I'm confused at what you assert such a program is doing up front. Turing presented a closed argument for ALL finite machines, which includes any machine and its languages built upon them. Prior to expecting me to look at such, I'd need to understand more about how you expect to find some/any alternative "halting decider" program that proves other "halting decider" programs used to discover whether you can list all programs that 'halt' has flaws without making those original ones work after all?

I think you might be misunderstanding the original arguments of incompleteness and the degree of their applications.

Another way of describing the problem is,

"Can you find an ideal single universal calculator that can solve all problems with complete satisfaction to solve all possible particular calculator-related problems there exists?"

Are you saying that you CAN, do not know, or cannot (but may have a disagreement about HOW the prior proofs have presented their case)?
I encoded the Peter Linz H_Hat in x86 machine language such that H_Hat correctly
decides halting on itself, something that Linz claims to be impossible.
Oh, okay. So it is a particular critique of his you are challenging. I can't speak on it without investing in reading and learning his own claims. I'll step aside given I can't argue for nor against what I don't understand.
Skepdick
Posts: 14366
Joined: Fri Jun 14, 2019 11:16 am

Re: Formalizing Natural Language Semantics

Post by Skepdick »

PeteOlcott wrote: Wed Mar 04, 2020 8:03 pm The set of all knowledge that can be fully expressed entirely on the basis of the meaning of its words.
Pete, this is utter gibberish.
PeteOlcott wrote: Wed Mar 04, 2020 8:03 pm excludes any aspects of sense data from the sense organs. is not circular at all
When you try to find a counter-example you fail because none exist.
Never mind a counter-example - I am trying to find a single example that fits your definition and I can't find any!
Not even one.

Words don't mean anything. Symbols don't mean anything. That is why we have to teach Turing machines to recognize them.
That is why we have to program the machine to understand the language!
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Skepdick wrote: Wed Mar 04, 2020 8:31 pm
PeteOlcott wrote: Wed Mar 04, 2020 8:03 pm The set of all knowledge that can be fully expressed entirely on the basis of the meaning of its words.
Pete, this is utter gibberish.
PeteOlcott wrote: Wed Mar 04, 2020 8:03 pm excludes any aspects of sense data from the sense organs. is not circular at all
When you try to find a counter-example you fail because none exist.
Never mind a counter-example - I am trying to find a single example that fits your definition and I can't find any!
Not even one.

Words don't mean anything. Symbols don't mean anything. That is why we have to teach Turing machines to recognize them.
That is why we have to program the machine to understand the language!
Language is the syntactic encoding of semantic meaning within the communication process:
http://blog.tnsemployeeinsights.com/wp- ... ocess3.png
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Scott Mayers wrote: Wed Mar 04, 2020 8:16 pm
PeteOlcott wrote: Wed Mar 04, 2020 8:08 pm
Scott Mayers wrote: Wed Mar 04, 2020 6:00 pm
My background is in machine languages. But I'm confused at what you assert such a program is doing up front. Turing presented a closed argument for ALL finite machines, which includes any machine and its languages built upon them. Prior to expecting me to look at such, I'd need to understand more about how you expect to find some/any alternative "halting decider" program that proves other "halting decider" programs used to discover whether you can list all programs that 'halt' has flaws without making those original ones work after all?

I think you might be misunderstanding the original arguments of incompleteness and the degree of their applications.

Another way of describing the problem is,

"Can you find an ideal single universal calculator that can solve all problems with complete satisfaction to solve all possible particular calculator-related problems there exists?"

Are you saying that you CAN, do not know, or cannot (but may have a disagreement about HOW the prior proofs have presented their case)?
I encoded the Peter Linz H_Hat in x86 machine language such that H_Hat correctly
decides halting on itself, something that Linz claims to be impossible.
Oh, okay. So it is a particular critique of his you are challenging. I can't speak on it without investing in reading and learning his own claims. I'll step aside given I can't argue for nor against what I don't understand.
It simply expresses the following generic halting problem proof counter-example as a state machine.
(1) H_Hat() cannot decide that itself would halt because this would cause itself to loop.
(2) H_Hat() cannot decide that itself does not halt because this causes itself to halt.
This boils down to the self contradictory semantics of the Liar Paradox.
Last edited by PeteOlcott on Thu Mar 05, 2020 6:34 am, edited 2 times in total.
Skepdick
Posts: 14366
Joined: Fri Jun 14, 2019 11:16 am

Re: Formalizing Natural Language Semantics

Post by Skepdick »

PeteOlcott wrote: Wed Mar 04, 2020 8:36 pm Language is the syntactic encoding of semantic meaning within the communication process:
http://blog.tnsemployeeinsights.com/wp- ... ocess3.png
Then tell me what meaning I have encoded in the bitstream 00100101010010010010010101
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Skepdick wrote: Wed Mar 04, 2020 8:45 pm
PeteOlcott wrote: Wed Mar 04, 2020 8:36 pm Language is the syntactic encoding of semantic meaning within the communication process:
http://blog.tnsemployeeinsights.com/wp- ... ocess3.png
Then tell me what meaning I have encoded in the bitstream 00100101010010010010010101
The only pre-established semantic decoding rules would indicate that it is no more than a string of binary digits.
Skepdick
Posts: 14366
Joined: Fri Jun 14, 2019 11:16 am

Re: Formalizing Natural Language Semantics

Post by Skepdick »

PeteOlcott wrote: Wed Mar 04, 2020 9:55 pm The only pre-established semantic decoding rules would indicate that it is no more than a string of binary digits.
What you have just done (inferring that the data represents a string of binary digits) is called type-inference.

It's an undecidable problem. It's not a string, it's a number. It's not binary, it's decimal. Logic/language is not explicit - it never will be.

https://en.wikipedia.org/wiki/Type_system#Type_checking
When a programming language evolves a more elaborate type system, it gains a more finely grained rule set than basic type checking, but this comes at a price when the type inferences (and other properties) become undecidable, and when more attention must be paid by the programmer to annotate code or to consider computer-related operations and functioning. It is challenging to find a sufficiently expressive type system that satisfies all programming practices in a type safe manner.
So unless your sender and receiver have agreed (a priori) on the encoding/decoding rules your game is doomed for failure.

The shit-show that is Philosophy is evidence of that. You are pre-supposing a shared/normalized language/grammar.
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Skepdick wrote: Wed Mar 04, 2020 9:57 pm
PeteOlcott wrote: Wed Mar 04, 2020 9:55 pm The only pre-established semantic decoding rules would indicate that it is no more than a string of binary digits.
What you have just done (inferring that the data represents a string of binary digits) is called type-inference.

It's an undecidable problem. It's not a string, it's a number. It's not binary, it's decimal. Logic/language is not explicit - it never will be.

https://en.wikipedia.org/wiki/Type_system#Type_checking
When a programming language evolves a more elaborate type system, it gains a more finely grained rule set than basic type checking, but this comes at a price when the type inferences (and other properties) become undecidable, and when more attention must be paid by the programmer to annotate code or to consider computer-related operations and functioning. It is challenging to find a sufficiently expressive type system that satisfies all programming practices in a type safe manner.
So unless your sender and receiver have agreed (a priori) on the encoding/decoding rules your game is doomed for failure.

The shit-show that is Philosophy is evidence of that. You are pre-supposing a shared/normalized language/grammar.
I am not pre-supposing anything. I am defining the architecture of a formal system capable
of rejecting finite strings that express semantically ill-formed truth-bearers.
Skepdick
Posts: 14366
Joined: Fri Jun 14, 2019 11:16 am

Re: Formalizing Natural Language Semantics

Post by Skepdick »

PeteOlcott wrote: Wed Mar 04, 2020 10:14 pm I am not pre-supposing anything. I am defining the architecture of a formal system capable
of rejecting finite strings that express semantically ill-formed truth-bearers.
OHHHHHH. You are building a syntax-checker!!! A parser/compiler/interpreter.

Pete, you can't be serious. Did you really spend 12000 hours on this?!?!?

Compiler theory is studied in any undergraduate computer science course.

Here is a free one: https://www.udacity.com/course/compiler ... ice--ud168
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Skepdick wrote: Thu Mar 05, 2020 8:48 am
PeteOlcott wrote: Wed Mar 04, 2020 10:14 pm I am not pre-supposing anything. I am defining the architecture of a formal system capable
of rejecting finite strings that express semantically ill-formed truth-bearers.
OHHHHHH. You are building a syntax-checker!!! A parser/compiler/interpreter.

Pete, you can't be serious. Did you really spend 12000 hours on this?!?!?

Compiler theory is studied in any undergraduate computer science course.

Here is a free one: https://www.udacity.com/course/compiler ... ice--ud168
https://en.wikipedia.org/wiki/Montague_grammar
It is the Montague grammar of natural language semantics.
Skepdick
Posts: 14366
Joined: Fri Jun 14, 2019 11:16 am

Re: Formalizing Natural Language Semantics

Post by Skepdick »

PeteOlcott wrote: Thu Mar 05, 2020 11:42 am https://en.wikipedia.org/wiki/Montague_grammar
It is the Montague grammar of natural language semantics.
The Montague grammar (or any grammar for that matter) is just one particular grammar in the set of ALL grammars.

Turing machines deal with ALL RECOGNIZABLE grammars. (the root of recognise is cognise... cognition)

If you can't teach a Turing machine to recognize it - it's not a grammar ( in the formal sense ).

Which is precisely the halting question/answer. Is X a grammar? Yes or No.

Given an English sentence, a Turing machine needs to be able to detect whether the sentence is grammatically correct..

Hard problem!
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Formalizing Natural Language Semantics

Post by PeteOlcott »

Skepdick wrote: Thu Mar 05, 2020 12:21 pm
PeteOlcott wrote: Thu Mar 05, 2020 11:42 am https://en.wikipedia.org/wiki/Montague_grammar
It is the Montague grammar of natural language semantics.
The Montague grammar (or any grammar for that matter) is just one particular grammar in the set of ALL grammars.

Turing machines deal with ALL RECOGNIZABLE grammars. (the root of recognise is cognise... cognition)

If you can't teach a Turing machine to recognize it - it's not a grammar ( in the formal sense ).

Which is precisely the halting question/answer. Is X a grammar? Yes or No.

Given an English sentence, a Turing machine needs to be able to detect whether the sentence is grammatically correct..

Hard problem!
https://www.geeksforgeeks.org/chomsky-h ... mputation/

There is a loophole in the halting problem proofs that no one ever noticed before.
Whether or not a universal halt decider can be created does not actually depend
on deciding whether the Liar Paradox is true or false.
Post Reply