DAML Scribblings (original) (raw)

DAML Scribblings and @@Fodder

@@@@@@@@@@

for the syntax section:

see also XML Schema for RDF, rdfs, inference, lists, problem report of 28Jul2000

@@stuff to be reviewed...

The RDF model and syntax specification defined a number of features which are not required for DAML. DAML agents will not be required to be able to process these.

RDF "aboutEachPrefix"

RDF bags (??@@) and "aboutEach"

xml:lang attributes are not represented in the RDF model

RDF documents are encoded in XML syntax [XML1.0] using namespaces [NS]. For the purposes of this project, the following XML features are not required for DAML, and DAML agents are not required to process them.

DTD parsing

DAML project participants should note that the current RDF-Schema document [RDFS0] has dissimilar definitions of range and domain. This document assumes that this error will be fixed and the conventional meanings_[@@what are those? --DanC 31Jul]_ associated in a future version of the specification [do the new definitions get new names, i.e. new URIs? --DanC 31Jul].

RDF within other XML documents
Practical considerations require DAML to be able to be inserted into other XML documents [which practical considerations? do we really need this now? --DanC 31Jul]. For this to be possible, the DAML parser must have a hint as to whether the other namesapces are ignorable, or have some RDF semantics. The document management section includes an RDF property for asserting that a given namespace is transparent to RDF in that any non-RDF elements may be replcaed by their content with no change to the RDF model conveyed by the document.

@@ check we have covered:

RDF Normal form
- DAML involves exchange of information reduced to RDF-NF
- Bits of RDF M&S+rdf-schema which you can ignore (foreach*...)
- Bits of XML which you can ignore (DTDs... see Ora)
- Semantics of RDF message = conjunction of statements which are assertions -
  semantics (stefan)
  * deleting arcs (@@ref earlier discussion)
  * reification will save you
  * namespace URI -> defining ontology (language) definition dociument

@@@@@@@@@@

from the Model section:

Abstract Algebra

(borrowing from the terminology of a section of the KIF 3.0 spec)

@@hmm... we could define "reflexive on S" rather than just "reflexive".

Property
- Reflexive
  p(x, x)
  if
  * type(p,Reflexive)
  * p(x, y)
  * Equivalence
- Transitive
  p(x,z)
  if
  * type(p,Transitive)
  * p(x,y)
  * p(y, z)
  * Equivalence
- Symmetric
  (commutative?)
  p(y,x)
  if
  * type(p,Symmetric)
  * p(x,y)
  * Equivalence
- Equivalence
  type(p,Equivalence)
  if (@@and only if)
  * type(p,Reflexive)
  * type(p,Transitive)
  * type(p,Symmetric)
- Acyclic
  implies(true, false)
  if
  * type(p,Acyclic)
  * path(p,l)
  * first(l,a)
  * reverse(l,lrev)
  * first(lrev,a)
isDisjointFrom: Class -> Class
implies(true, false)
if
- isDisjointFrom(x,y)
- type(z,x)
- type(z,y)
path: Property -> List
path(p,l)
if
- first(l,a)
- rest(l,lrest)
- first(lrest,b)
- p(a,b)
- rest(lrest,empty)
  path(p,l)
  if
- first(l,a)
- rest(l,lrest)
- first(lrest,b)
- p(a,b)
- path(p,lrest)

Equivalence and derived classes

isEquivalentTo: Resource -> Resource@@links

The concept of equivalence is a crucuial one upon which much else is then based. It must be stressed that this is one equivalence relation which is defined here, while anyone is free to define their own for their own purposes. The one defined here is a strict higher-order equivalence, in that the semantics are that the resources connected by isEquivalentTo relation can be subsituted for each other under any circumstances.

The only processing which is disallowed (following web architecture) is that properties are not allowd to peek into the content of the URI string identifying a resource. That is, to define a property "The schema string of the URI identifying the resource" is illegal.

isEquivalentTo

isEquivalentTo(a,b) => ( a(x,y) => b(x,y)) & (x(a,y)=>x(b,y)) & ..

isInverseOf: Property -> Property
equivalent: see OIL's isInverse of
@@@ Question as to whether you count this as following isEquivalentTo or not. It gives you a way of expressing equivalence of properties as
inverse(a,b) and inverse(b,c) gives effect of equivalence.
q(y, x)
if
- isInverseOf(p,q)
- p(x, y)
  type(isInverseOf, Symmetric)

More types of Properties

Property
- Unambiguous
  aka Injective, or 1-1. Compary max-cardinality=1.
  isEquivalentTo(x,z)
  * type(p,Unambiguous)
  * p(x, y)
  * p(z, y)
- Unique
  aka functional
  isEquivalentTo(y,z)
  * type(p,Unique)
  * p(x, y)
  * p(x, z)
- Antisymmetric
  isEquivalentTo(x,y)
  if
  * type(p,Antisymmetric)
  * p(x, y)
  * p(y, x)
- PartialOrder
  transitive, antisymmetric@@

@@@ Check: Have we included

Some useful properties and classes for the above
"a survey of the literature"
split into things whioch require equivalence and things which do; things implying HOL; etc- in fact we have no FOL layer all is HOL.
look for dependencies elsewhere;
look for a table of what system understands what terms;
Write this in dan-ish
- Equivalence, Uniqueness (i.e. max. cardinality=1) F => s/x/y/ (F)
- inverse -- type-> property
- acyclic, symmetric, reflexive, daml:transitiveThing -- type -> class; -- subclass -> property
- rdfExtractor -- type -> property
- mutuallyDisjoint
- Containers? List, enumerated set (Stefan, Deborah)
- @@@@ contributions please
- Chapters 2 and 1 and 5 define DAML-0.5 deliverable
- Note on how you represent nary predicates in DAML
- Examples of shoe rules expresses in DAML-0.5
- Check staphan stchabe's articel @@link;
- @@ People check Ontolingua (Stefan), SHOE (Jeff); KRSS (Lynn); Netscape MCF submission (DanBri)
  OKBC Knowledge Model @@GFP knowledge model
- Discuss datatyping with respect to XML Schema stuff if they have URIs
  In contrast to HTML - if you want to change this list you don't change this list but you put something different ideas on the web. Use equivalence to show when yours is the same as mine. Future standard may be a new list with a lot of this list and others.
- Examples of rules in RDF syntax. Circles and arrows.

@@Building HOL section

@ disclaimers of status of thius chapter

@@ Dan to put his work on this here

@@@ check have we included:

Building HOL on top of RDF-NF
- Really a second chapter 2 - not dependent on ch 2
- HOL axioms (and, implies, false, forall?)
- The arbitrariness of the choice of axioms
- Derived useful things
- Explaining how to write rules
- Lots of Dan-derivation and examples
- Pointer to test datasets
- Examples of SHOE rules (etc) reduced to FOPC primitives

Solidifying DAML 0.5

This chapter is not formally part of DAML as it assumes conversance with the previous chapter in which logical primitives were introduced. Here we use these primitives to give a formal set of rules for DAML; rules which were implicitly or explicitly stated in the RDF Model and Syntax, RDF Schema documents, and in the Chapter on ontological terms in this document.

RDF Model and Syntax, RDF Schema

Ontological Rules

The raw transfer syntax RDF corresponding to these rules is given in an appendix.

@@ from Doc Mgmt section:

Housekeeping terms

Supercedes

Supercedes: Class -> Class
implies(true, false)
if
- isDisjointFrom(x,y)
- type(z,x)
- type(z,y)

@@ check:

Practicalities of logic on the web: document management etc
- supercedes domain-> namespace; range->namesapce, (obsoletes?); derived-from?
- is-time-specific-version-of
- persistent -- rdf:type-> class; --superclass-> nameSpace will not change
- sublanguage (relationship between namespaces)
- toolbox bits such as toolbox:include, according-to etc
- this ontology includes this other ontology
- deprocates domain->property; range->property [Jeff suggets useful]
- rdf-extractor maps document (or XML namespace?) to script.
- 'According to' (says, justification) doc => F

@@@@@@@@

A gloss on the RDF spec, in KIF terms:

resource

object (though I think the RDF spec confuses resources with URIs in some cases)

URI*

word (this is a rough correspondence)

*absolute URI with optional fragment identifier, that is

statement

sentence, or relational sentence

property

relation

predicate

relation constant

subject

term; in particular: object constant

object

term; in particular: object constant

@@@@Future Work

@@check:

inconsistency
trust
- DSIG
use cases
- cf pc week article
- target audience
internationalization
order
english equivalence and GUI aids
(authoring tools --Dan)
graphical notation
whiteboard ASCII syntax??? Notation3? Metalog? namespace-kif? (DC against?)
Efficient transmission syntx for reified statements?
temporal logic
Ontology (if danbri can define it)

@@@

auto generating stuff...

(@@hmm... relative URIs aren't handled correctly)

./DAML-0-5.html,apply=/magic/dan5.xsl

(to allow relative URI for target of conversion!)

@@@@@@

some stuff in bibliography of proposal