Use Cases: two examples of use cases

Here are two examples of partly-made-up use cases:



Name:  Propositional Attitudes
Submitted by:  Peter F. Patel-Schneider
WG supporter:  Peter F. Patel-Schneider
Status: Submitted, supported, complete, not approved


Scope:

I want to be able to represent beliefs, and related propositional attitudes 
such as imagine and consider.  Know is somewhat similar but not included in 
this use case.  Beliefs and other propositional attitudes are between a person 
(considered generally, i.e., not requiring that the person be a human) and a 
triple.


Details:

Propositional attitudes would be modelled as a relationship between a person 
and a quoted triple, as in:

:John :believes << :Lois :loves :Superman >> .
:John :believes << :Clark :loves :Lois >> .


Propositional attitudes do not require that the triple be asserted, so that:

:John :imagines << :Lois :loves :Superman >> .

does not entail

:Lois :loves :Superman .

and SPARQL returns no results for

SELECT ?lover ?lovee WHERE { ?lover :loves ?lovee }


Because propositional attitudes do not admit substitution of equals, this 
requires that IRIs in them be opaque, so that

:John :believes << :Lois :loves :Superman >> .
:Superman owl:sameA :Clark .

does not entail

:John :believes << :Lois :loves :Clark >> .


Propositional attitudes should admit substitutions of equal literals, because 
any person considers two equal literals to be the same, so that

:John :guesses << :Lois :age "42"^^xsd:int >> .

entails

:John :guesses << :Lois :age "042"^^xsd:int >> .

entails

:John :guesses << :Lois :age "42"^^xsd:integer >> .


It is unclear what the situation should be for blank nodes in beliefs.  Consider

:John :believes << :Lois :loves _:x >> .

One reading is that it models the situation where John believes that Lois 
loves someone but does not know who that someone is.  Another reading is that 
is models the situation where John believes that there is a particular person 
whom Lois loves.  The problem is that RDF has no way of distinguishing between 
the two sitations because it has no way to specify quantifier scope.


Propositional attitudes are between a person and an abstract triple not an 
occurrence of a triple, i.e., if two people believe triples with the same 
components then they believe the same thing, so

:John :believes << :Lois :loves :Superman >> .
:Jack :believes << :Lois :loves :Superman >> .

entails

:John :believes _:x .
:Jack :believes _:x .

and

:John :believes << :Lois :age "42"^^xsd:int > .
:Jack :imagines << :Lois :age "42"^^xsd:integer > .

entails

:John :believes _:x .
:Jack :imagines _:x .


So propositional attitudes require that quoted triples be unique when the have 
identical subjects, predicates, and objects or when they have identical 
subjects and predicates and equal literal objects.



Need for quoted triples:

Propositional attitudes need to reference triples, which is one of the main 
reasons that quoted triples exist.  As with any use of quoted triples RDF 
reification or a similar method could be used.  Quoted triples have the 
advantage of uniqueness so that there is no need to ensure that the same 
center node is used for all reifications of the same triple. There is also the 
inherent efficiency advantage and representational advantages of having a 
simpler construct instead of the unwieldly RDF reification of a triple.


Adequacy of quoted triples:

Propositional attitudes towards a single atomic proposition are completely 
handled by quoted triples.


Commentary:

The working group agrees that quoted triples of this sort are adequate to 
handle propositional attitudes to single atomic propositions.  But 
propositional attitutes are not to single atomic propositions but instead to 
sets of atomic propositions or entire belief states.  The former can be much 
better represented as an entire RDF graph and named graphs used for 
propositional attitudes.  The latter requires facilities beyond those provided 
by RDF.

The working group thus is not approving this use case as one that will 
necessarily be supported by the recommendations it produces.






Name:  Wikidata
Submitted by:  ??
Status: Submitted, incomplete


Scope:

As a member of the Wikidata community,
I would like to see triple stores supporting the Wikidata/Wikibase data model 
as much as possible, so that provenance information etc. can be represented in 
a way which is pleasing to the mind and software-systems such as Wikibase.  
For further information see:

https://www.mediawiki.org/wiki/Wikibase/DataModel
https://www.mediawiki.org/wiki/Wikibase/Indexing/RDF_Dump_Format
https://phabricator.wikimedia.org/T206560

Commentary:

This use case needs many more details.  The Wikidata data model has many 
aspects whose ideal translation into RDF is uncertain even with the RDF dump 
format providing one translation into RDF.  What aspects of the Wikidata data 
model do you want covered and how are they to be covered using quoted 
triples?  Without this information the working group cannot further consider 
this use case.