perm filename RLL.MSC[RLL,DBL] blob sn#677959 filedate 1982-09-17 generic text, type C, neo UTF8
COMMENT ⊗   VALID 00005 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002		List of all sentences
C00015 00003		Predicate Calculus Statements for Various RLL things
C00022 00004		What's in a Language?
C00027 00005	Questions re: RLL
C00032 ENDMK
C⊗;
	List of all sentences
  1. All purple mushrooms are poisonous.
  2. Everybody loves somebody. {Two meanings}
  3. Purple mushrooms have many attributes.
  4. There are 23 purple mushrooms in the world.
  5. Purple-mushroom-23 has been accessed 12 times.
  6. The fact that a purple mushroom is purple is definitional.
  7. The integers mod 5 form a group under addtion.
  8. John gave Mary the book.
  9. Apple-23 embodies the {theory | concept} pf apple. [is an examplar]
  10. The color of all singleton sets is red.
↓ 11. The alleged theif hated the poor violinist.
  12. Most elephants are grey.
↓ 13. John poured some of the liquid out of the bottle.
? 14. Tower of Hanoi

!Need: 
4 forms of Isa:
  (1) Standard - NON-Definitional
	Expresses merely relation of set membership
  (2) Existential  - for ∃xεS 
  (3) Universal    - for ∀xεS 


New Any$SELF$Slots:  DefinitionalSlots and AttributeSlots
This may require splitting many slots into multiple slots.

Conventions:
Any---  is the set of all ---s
Typical--- is a typical element of this set of ---

Every Typical--- unit has a slot InheritableSlots, which points to slots...
	also DefinitionalSlots

If sεAny$SELF$Slot, 

TypicalDog
	TypicalExampleOf:	AnyDog
	InheritableSlots:	(Parts, ..., MyWorth)
	Parts:			(4 legs, ...)
	MyWorth:		45
	MyWorthForInstance:	33

MyWorth
	Isa:			(Any$SELF$Slot)
	StoredInTypAs:		MyWorthForInstance

Now when Dog#53 is initialized,
its MyWorth value with be 33, NOT 45 (45 is worth of TypicalDog, qua unit.)

** Note ***
Slot1
	SuperSlot:	SSlot

means
1. Domain(Slot1)=Domain(SSlot)
2. ∀xεUnit. Slot1(x) ⊂ SSlot(x)

Typical1
	SuperTypEx:	STypical

means
TypicalExampleOf( Typical1 ) ⊂ TypicalExampleOf( STypical )
!1. All purple mushrooms are poisonous.

   ***	SOLN#1  ***

AnyMushroom
	SubClass: AnyPurpleMushroom

AnyPurpleThing
	SubClass: AnyPurpleMushroom

AnyPurpleMushroom
	TypicalExample: TypicalPurpleMushroom

TypicalPurpleMushroom
	Poisonous: T
	InheritableSlots: (Poisonous)

   ***	SOLN#2  ***

AnyMushroom
	UniversalExamples: {x ... }

x
	UniversalIsa:	{AnyMushroom}
	Color:		Purple
	Poisonous:	T
	DefinitionalSlots: {UniversalIsa, Color}
	InheritableSlots: (Poisonous)

Isa
	Isa:		{Slot}
	SubSlots:	{ExistentialIsa, UniversalIsa, StandardIsa}

UniversalIsa
	Isa:		{Slot}
	SuperSlots:	{Isa}

   ***	SOLN#3  ***

AnyMushroom
	UniversalExamples: {x ... }

AnyPoisonousThing
	StdExamples: {x ... }

x
	UniversalIsa:	{AnyMushroom}
	StdIsa:		{AnyPosionousThing}
	Color:		Purple
	DefinitionalSlots: {UniversalIsa, Color}
	InheritableSlots: (Poisonous)

TROUBLE - unnatural!!!

   ***	SOLN#4  ***

AnyMushroom
	SubClass: {x ... }

X
	SuperClass: {AnyMushroom}
	TypicalExample: TypicalX
	DefinitionalSlots: {SuperClass, TypicalExample}

TypicalX
	TypicalExampleOf: {X}
	Poisonous:	   T
	Color:		Purple
?	DefinitionalSlots: {?, Color}
	InheritableSlots: (Poisonous)

!2a. Everybody loves somebody. ∀xεPerson ∃yεPerson. Loves(x,y)

AnyPerson
	UniversalExamples: {x, y...}

x
	UniversalIsa:	{AnyPerson}
	Loves:		y
	DefinitionalSlots: {UniversalIsa}
	VariesThru:	AnyPerson

y
	ExistentialIsa:	{AnyPerson}
	LovedBy:	x
	DefinitionalSlots: {LovedBy, ExistentialIsa}
	VariesWith:	x
!2b. Everybody loves somebody. ∃yεPerson ∀xεPerson. Loves(x,y)

   ***	SOLN#1  ***

AnyPerson
	TypicalExample:  TypicalPerson
	UniversalExamples: {y, ...}

TypicalPerson
	TypicalExampleOf: AnyPerson
	Loves:		  y

y
	ExistentialIsa:	{AnyPerson}
	LovesBy:	TypicalPerson
	DefinitionalSlots: {LovedBy, ExistentialIsa}
	VariesWith:	NIL

   ***	SOLN#2  ***

AnyPerson
	UniversalExamples: {x, y...}

x
	UniversalIsa: {AnyPerson}
	Loves:		  y
	DefinitionalSlots: {UniversalIsa}
	VariesThru:	AnyPerson

y
	ExistentialIsa:	{AnyPerson}
	LovesBy:	TypicalPerson
	DefinitionalSlots: {LovedBy, ExistentialIsa}
	VariesWith:	NIL
!3. Purple mushrooms have many attributes.

AnyMushroom
	UniversalExamples: {x ... }

x
	UniversalIsa:	{AnyMushroom}
	Color:		Purple
	Poisonous:	T
	DefinitionalSlots: {UniversalIsa, Color}
	#Slots:		Many

#Slots:
	Isa:	{Any$SELF$Slot}

!4. There are 23 purple mushrooms in the world.

AnyMushroom
	Cardinality:	23

Cardinality
	Isa:	{Slot}

!5. Purple-mushroom-23 has been accessed 12 times.

AnyMushroom
	Examples: {Purple-mushroom-23, ...}

Purple-mushroom-23
	Isa:	{AnyMushroom}
	#Accesses:	12

#Accesses
	Isa:	{Any$SELF$Slot}

!6. The fact that a purple mushroom is purple is definitional.

AnyMushroom
	UnverisalExamples: {x ... }

x
	UniversalIsa:	{AnyMushroom}
	Color:		Purple
	Poisonous:	T
	DefinitionalSlots: {UniversalIsa, Color}
	VariesThru:	AnyMushroom
!7. Block1 and Block2 together form a blivit.
   Block2 and Block3 together form a blivit.
   Block1 and Block3 together form a blivit.

AnyBlivit
	EssentialParts:	{Whiggit, Whatever}
	Examples:	{Blivit-0, Blivit-32, Blivit-95 ...}

Blivit-0
	Isa:		(AnyBlivit)
	Whiggit:	Block1
	Whatever:	Block2

Blivit-32
	Isa:		(AnyBlivit)
	Whiggit:	Block2
	Whatever:	Block3
	APartOf:	Block3

Blivit-95
	Isa:		(AnyBlivit)
	Whiggit:	Block1
	Whatever:	Block3

Block1
	Isa:		{AnyBlock, AnyWhiggit}
	WhiggitFor:	(Blivit-0, Blivit-32)
	APartOf:	(Blivit-0, Blivit-32)

Block2
	Isa:		{AnyBlock, AnyWhiggit, AnyWhatever}
	WhateverFor:	(Blivit-0)
	WhiggitFor:	(Blivit-32)
	APartOf:	(Blivit-0, Blivit-32)

Block3
	Isa:		{AnyBlock, AnyWhatever}
	WhiggitFor:	(Blivit-32, Blivit-95)
	APartOf:	(Blivit-32, Blivit-95)

APartOf
	Isa:		(AnySlot)
	SubSlots:	(WhiggitFor, WhateverFor)

Whiggit
	Isa:		(AnySlot)
	Inverse:	WhiggitFor

Whatever
	Isa:		(AnySlot)
	Inverse:	WhateverFor

WhiggitFor
	Isa:		(AnySlot)
	SuperSlots:	(APartIn)
	Inverse:	Whiggit

WhateverFor
	Isa:		(AnySlot)
	SuperSlots:	(APartIn)
	Inverse:	Whatever
!8. John gave Mary the book.

AnyGivingEvent
	Examples:	{GE-41, ...}

GE-41
	Giver:		John
	Recipient:	Mary
	Object:		Book

John
	GiverIn:	GE-41

Mary
	RecipientIn:	GE-41

Book
	ObjectIn:	GE-41

GiverIn
	Isa:		{Slot}
	SuperSlots:	{ParticipatesIn}

RecipientIn
	Isa:		{Slot}
	SuperSlots:	{ParticipatesIn}

ObjectIn
	Isa:		{Slot}
	SuperSlots:	{ParticipatesIn}
!9. Apple-23 embodies the {theory | concept} pf apple. [is an examplar]

AnyApple
	Examplar:	Apple-23

Apple-23
	ExamplarOf:	AnyApple

Examplar
	Isa:	{Slot}
	?

x = Examplar( S ) means:

1. x ε S
2. ∀yεS. ∀sεT.  s(y) = s(x) most of the time.
	where T = InheritableSlots(TypicalExample(S)).
	[Rewording: ∀sεT. |{yεS | s(y) = s(x)}| ≥ 80% of |S| ]
!10. The color of all singleton sets is red.

AnySet
	UniversalExamples:	{x, ...}

x
	UniversalIsa:	{AnySet}
	Cardinality:	1
	DefinitionalSlots: {UniversalIsa, Cardinality}
	Color:		Red
!12. Most elephants are grey.

AnyElephant
	TypicalExample:	TypicalElephant

TypicalElephant
	Propositions:	{Prop-41, ...}

Grey
	Isa:		{AnyColor}
	Propositions:	{Prop-41, ...}

Prop-41
	Relation:	Color
	Value:		Grey
	Modality:	MostAre
	FromUnit:	TypicalElephant

Propositions
	Isa:		{Slot}
	Datatype:	... (*P AnyProp) ...

AnyProp
	TypicalExample:	TypicalProp

TypicalProp
	TypicalExampleOf:	AnyProp
	SensibleSlots:		{Relation, Value, FromUnit, Modality, Tense, ...}
!-- Or --

AnyElephant
	TypicalExample:	TypicalElephant

TypicalElephant
	Color:	{(Prop Prop-41), ...}

Grey
	Isa:		{AnyColor}
	ColorOf:	{(Prop Prop-41), ...}

Prop-41
	Relation:	Color
	Value:		Grey
	Modality:	MostAre
	FromUnit:	TypicalElephant

!-- OR --

AnyElephant
	UniversalExamples:	{x, ...}

x
	UniversalIsa:		{AnyElephant}
	QualifiedBy:		Most
	DefinitionalSlots:	{UniversalIsa, NatureOfQualification}
	Color:			Grey

Most
	?

--------------

AnyElephant
	UniversalExamples:	{x, ...}

AnyGreyThing
	StdExamples:		{x, ...}

x
	UniversalIsa:		{AnyElephant}
	QualifiedBy:		Most
	DefinitionalSlots:	{UniveralIsa, QualifiedBy}
	StdIsa:			{AnyGreyThing}
	Isa:			{AnyGreyThing, AnyElephant}
!14. Tower of Hanoi
??

	Predicate Calculus Statements for Various RLL things
(15-Oct-80)
Typed logic
  u, u-i, u0, u1, u2, ...	Satisfy Unitp, i.e. represents Units
  s, s-i, s0, s1, s2, ...	Satisfy Slotp, i.e. represents Slots 
				(subtype of Units)
  f, f-i, f0, f1, f2, ...	represents Functions
  l, l-i, l0, l1, l2, ...	Satisfy LISTP, i.e. represents Lists
  t, t-i, t0, t1, t2, ...	represents Times
[ p's are used for pseudo-slots ]

	LISP OVERHEAD:
	[Note- this is essentially MEMB]
∀ v,l. v in l ≡ [ (l ≠ NIL) & ( v = CAR(l) ∨ v in CDR(l) )]

	Slot Related Overhead:

∀ s. HighLevelDefn( s ) = GetValue( s,'HighLevelDefn )
∀ s. Format( s ) = GetValue( s,'Format )
∀ s. Defn( s ) = GetValue( s,'Defn )
∀ s. Domain( s ) = GetValue( s,'Domain )
∀ s. Range( s ) = GetValue( s,'Range )
∀ u,s. GetValue-1( u,s ) = GetValue( u,s )
∀ u,s. GetValue( u,s ) = OR( UA-GETVALUE( u,s ), APPLY( Defn( s ), u ) )

∀ p. Format( p ) = CAR( Range( p ))
∀ p,HLD. HighLevelDefn( p ) = HLD ⊃
  Parse( HLD ) = LIST( Defn( p ), Range( p ), Domain( p ), ? ).

∀ u,p. GetValue-1( u,p ) = APPLY( Defn( p ),u ).

∀ u,p,v. InValue( u,p,v ) ≡
  { [ (Format( p ) ε {FSingleton} ) & v = GetValue-1( u,p ) ]
    [ (Format( p ) ε {FSet, FList} ) & v in GetValue-1( u,p ) ] }

	CONSISTENCY:
∀ u,s,v,v',t0,t1,t2. 
  [ UA-PUTVALUE( u,s,v )(t0) & [t0<t1<t2 ⊃ ¬UA-PUTVALUE( u,s,v' )(t1) ] ] ⊃
	UA-GETVALUE( u,s )(t2) = v.

∀ u,s,v,v',t0,t1,t2. 
  [ PutValue( u,s,v )(t0) & [t0<t1<t2 ⊃ ¬PutValue( u,s,v' )(t1) ] ] ⊃
	GetValue( u,s )(t2) = v.

∀ s1,s2. s2 = GetValue( s1,'Inverse ) ⊃
  [∀ u,v. InValue( u1,s,v ) ⊃ [Unitp( v ) & InValue( v,s2,u )] ]

SLOT DEFINATION:

	COMPOSITION:
∀ p1,HLD,u. HLD = HighLevelDefn( p1 ) & Composition = CAR( HLD ) ⊃
  ∃ p2. HighLevelDefn( p2 ) = CAR (CDR ( HLD ) ) &
  [IF CDR( CDR( HLD ) ) = NIL
     THEN GetValue-1( u,p1 ) = GetValue-1( u,p2 )
     ELSE ∃ p3. HighLevelDefn( p3 ) = CONS('Composition, CDR( CDR (HLD) ) )] &
	    ∀ v. InValue( u,p2,v ) ⊃
		Unitp( v ) & GetValue-1( v,p3 ) = GetValue-1( u,p1 )

	UNIONING:
∀ p1,HLD,u. HLD = HighLevelDefn( p1 ) & Composition = CAR( HLD ) ⊃
  [∃ p2. HighLevelDefn( p2 ) = CAR (CDR ( HLD ) ) &
	∀ v. InValue( u,p2,v ) ⊃ InValue( u,p1,v )] &
  [IF CDR( CDR( HLD ) ) ≠ NIL
     THEN ∃ p3. HighLevelDefn( p3 ) = CONS('Unioning, CDR( CDR (HLD) ) )] &
	    ∀ v. InValue( u,p2,v ) ⊃ InValue( u,p1,v )

	STARRING:
∀ p1,HLD,u. HLD = HighLevelDefn( p1 ) & Starring = CAR( HLD ) ⊃
  InValue( u,p2,u ) &
  ∃ p2. HighLevelDefn( p2 ) = CAR (CDR ( HLD ) ) &
     ∀ v,v1. InValue( u,p2,v ) ⊃
	Unitp( v ) & [InValue( v,p1,v1 ) ⊃ InValue( u,p1,v1 ) ]

!TO CSD.GENESERETH, CSD.SMITH 13:38 17-June
Proposition:

	RLL will come with a large (and extendable) set of Simulation Structures.
The initial ("top level") SS (that is, the starting setting for the
organ,) defines each SS as a set of linguistic elements (which are the set of
LISP functions,) only some of which have a semantic interpretation. That subset
is different for each SS; as is the actual interpretation.
(Hence, RDG's RLL has defined GetValue and PutValue, whereas MRG's version makes
no effort to define these, and instead worries about True and Value.)

	These functions are used to access the information in the KB;
each SS somehow states that all the other symbols of LISP are
available to the user, but does not provide these linguistic elements
with any (pre-defined) semantics. 

	When creating a new KB, the user may
(1) indicate which SS he wishes
(2) indicate this KB will have a new type of SS - and will proceed to 
	to define it (using mechanisms of the current SS.)

Questions:
1) Does this seem feasible? Is it basically what you had in mind? What would you
	want to change?

2) Should each KB have exactly 1 Simulation Structure associated with it?
	Is this a limitation - when might a user wish a single KB to have 
	more than one SS?
3) How many functions should there be for accessing the KB? I believe this
	number should be small, and these functions should have a good "semantics".

Comments in general.
	What's in a Language?
∂TO CSD.SMITH, CSD.GENESERETH, STT (?/80)
List of terms to be defined -- or at least have a common
meaning to those of us here:

Language
	A set of words and a grammar.  Note this is a-semantic.
	They are inherently Linguistic entitites -- in the manner of 
	a grammar is. [See RWW's LS-pairs, ie Language/Structure]
Representation
	A mapping between words in a language and objects,
	relationships, etc. in the (some) world.
	Representation is both an interpretation and a set of symbols
	In this sense, language is a component of a representation
	[See SIGART, p. 68-9 for defn, p.5 for illustration of difficulties]
Representation Language
	Corresponds to "real world" IN NATURAL WAY -
	so the pieces are easy to manipulate
	Each is a scheme for constructing and augmenting
	a representation, including tools for facilitating this process.
	Note than a representation language IS A representation and a language.
Representational Components
	Pieces used in a representation languages.
Representation Language Language
	A scheme for constructing and augmenting
	a representation language, including tools for facilitating same.
	Note than a representation language language IS A representation language.
Linguistic
Epistemologic
	(Epistemological neutrality)
Semantics
Syntax
Model
Interpretation
Frame/Schema/Unit
Slot
Proposition
Inheritance
Default
Modality
Object Centered
Function Centered
Stream Centered
"Reason maintanence" (nee "Truth maintanence")
Justification
Meta-level
Instance
Perspective/View
Self-descriptive
Self-modifiable

Specific to RLL
---------------
Slot Type
Class of Units
Typical Member

!∂07-Oct-80  0742	CSD.SMITH at SU-SCORE 	Re: Start of Glossary      
Date:  7 Oct 1980 0736-PDT
From: CSD.SMITH at SU-SCORE
Subject: Re: Start of Glossary  
To: RDG at SU-AI
In-Reply-To: Your message of 6-Oct-80 1659-PDT

RPLACA(("Truth maintenance" ...) "Reason maintenance"]
-------

Date: 27 Oct 1980 1217-PST
From: CSD.SMITH
Subject: next meeting
To: CSD.GENESERETH, CSD.SMITH, CSD.GREINER, CSD.STT, csd.bennett, JD at SU-AI,
    LGC at SU-AI, CSD.CPP, RMS at SU-AI

For this fridays representation meeting the subject will be SDL and FOL.
Readings are Weyrauch's PROLEGOMENA   (in recent AI Journal issue)
	and  Dolyle's SDL chapter (chapter 2).
Also you might think about the following questions (which I hope we can reach
a consensus on):

	What is a "representation"?
	What is a "theory"?  Are they equivalent?
	What does it mean to be an "object" or "individual"?
		(If you think this question is easy, think about your desk or
		office as being a quantum-mechanical wave function.)
	What is the relationship between theories and contexts?
	What would it mean for a theory to refer to itself?
	What would it mean for a process to (have a theory of and) change itself
	...etc.

Cheers.  Dave.
-------
                ---------------

-------

Questions re: RLL

 1-Aug-80 13:16:09-PDT,0000000708;000000000001
Date:  1 Aug 1980 1316-PDT
From: CSD.HINES
Subject: question on rll mechanism
To: csd.greiner

Russ,
   Can or how does one refer to slots or units inside LISP code?
What I mean is in writing a resolving function, I would like to refer
to a few slots which may need to be searched for via Isa's links.
Example:  A resolving function needs to know if the two clauses that
it has picked out have been previosly resolved together in the same 
manner.  To do this I plan to create the slot under ClauseSet called
Resolewe which will contained a list of (clause sigma) where clause is the
name of a clause which has been resolved with the queried clause and sigma
is the unification used.
-------
13-Nov-80 10:31:23-PST,00000002194;000000000001
Date: 13 Nov 1980 1029-PST
From: CSD.HINES
Subject: πGetValue/ToGetValue/Epistemological paradoxes
To: csd.greiner

     In reading through Appendix A of RLL, I seem to have tied 
myself in a knot.  On page 27, line 5-15, does GetValue actually
cycle through l6-10 until it comes to (GetValue 'Togetvalue 'Togetvalue)
at which time it 'knows' to stop this recursion and actually look at
the value of ToGetValue:ToGetValue, GetAccessFn?  This would seem to
be the wrong approach for 2 reasons:

        1)  It's artificial (Maybe this is O.K. since it is Artificial
	    Intelligence).  I don't require knowledge about how my brain
	    retrieves information before I can actually retrieve it.  The 
	    The knowledge about retrieval (i.e what would be the value of
            ToGetValue:ToGetValue) is not crucial to the process itself.
	    [It is only important when modification is needed which presumably
	     is done by an entirely different agent. I don't even want to
             start thinking about a "self-modifiable" system.]

        2)  It's circular.  At some point GetValue must stop asking the 
            question "How do I get this Value" and must actually go to 
	    the slot: ToGetValue of the Unit: ToGetValue and find this
            fn.: GetAccessFn which tells it (I think) how to get a value.
	    I.E. To answer the question "How do I get a value?" it goes to
	    the slot which says "Among other things, You go to slots to see
	    their value." To this RLL replies "I see, I do what I just did,
            right?"  In other words, going to te value of ToGetValue
            presupposes knowledge that it suupposedly is garnering by going
	    there.

    So, I will assume (until 3:00 today when I will hopefully see you) that
 GetValue is "hardwired" in some way and that  while knowledge exists about
 how it works (if nowhere else in your head) this knowledge is not 
 necessary to its functioning.  This seems to indicate that RLL could
 only be modified by an external agent.
   Does any of this make sense. If not, at least you'll be a little
   better prepared to answer  my questions this afternoon.
-------
-------