An RDF Model for GET/PUT and Document Management

Assertion by reference

cfAssertion of another documentin thesemantic web toolbox.

In the general case, a reference to a term from another document in the Web does not imply all the assertions made in that document; for example, one document might refer to another for the purpose of refuting it. But for the case that we do want to assert the contents of another document, we have:

• hasBasis: Resource -> Resource

  for hasBasis(this, that) read: that is a basis for this; i.e. this asserts that by reference; i.e. this says/implies everything that says/implies. wordnet defn: "a relation that provides the foundation for something". @@subproperty of rdfs:seeAlso:

The shoe uses-ontology relation[@@link] is a subProperty of basis, for example.

To define basis, consider the rule "if r2 is a basis for r1, and r2 says s, then r1 says s". That's a reasonable rule, assuming the Web is static, i.e. that if some document ever makes a claim, it always makes that claim. But it does not allow us to model a page about a group that says who the membership of the group is, where the membership can change from time to time.

So rather than "r1 says s" we use "m1 says s" where m1 is a particular message that exposes the content of r1 -- an HTTP reply, for example:

consequent(a1, s)

if

• answer(q1, a1)
• subject(q1, r1)
• consequent(a1, basis_r1_r2)
• rdf:subject(basis_r1_r2, r1)
• rdf:predicate(basis_r1_r2, hasBasis)
• rdf:object(basis_r1_r2, r2)
• answer(q2, a2)
• subject(q2, r2)
• precedes(q2, a1) @@not sure this is quite right; do I need this? do I want some sort of freshness test in stead?
• consequent(a2, s)

This rule is written in terms of a model for communications protocols.

Communication Protocol Semantics

cf notes on state; no longer trying to stick to Singular Nouns for Names

an interesting exception (found5Jan2001):

Any software licences or redistribution permissions, release status documentation, and the like you get via CVS are NOT INTENDED to have legal effect. Just because I check a document into CVS does not mean that I stand by its contents.

-- Legal notices, release status, etc in ijackson's stuff

• entails: Message -> Statement

  for consequent(m, s) read s is a consequent of m; or: the intent of m is to assert (among other things) s. compare with "gate" symbol in modal logic? cf BAN logic@@, PCA@@. For example, if an HTTP response has status 200 OK and RDF/XML content, and some statement s is in the model expressed by that RDF/XML content, then consequent(m, s). Hmm... consequent(m, s1) /\ implies(s1, s2) => consequent(m, s2).

• hasAnswer: Message -> Message

  for hasAanswer(q, a) read: a is an (the?) answer to q. For example, q might be an HTTP GET request (or a GFP ask request), and a its answer.

  precedes(q, a)

  if

  • hasAnswer(q, a)

• precedes: Message -> Message

  for precedes(m1, m2) read: m1 precedes m2, i.e. m2 happens before m1, i.e. m1 <= m2.

  type(precedes, PartialOrder).

• subject: Message -> Resource

  for subject(m, r) read: m has subject r; i.e. the subject of m is r. e.g. if you do an HTTP GET for u1, then the subject of that GET message (and its response, I suppose) is u1.