This is my proposal to extend the Subject Predicate Object (SPO) triple model
of RDF to address some "deficiencies".
Intead of a ( S P O ) triple, I propose to use a tuple:
( S P O R C A E DNT AD GS IW IW2 )
S Subject (as in RDF)
P Predicate (as in RDF)
O Object (as in RDF)
R Responsible person (URI)
The URI of the person that will be held accountable for this information
(no blank nodes or literals). For information generated by reasoners:
The person who installed the reasoner. Changing any value in a tuple
requires R to be changed to the person responsible for the change.
C Context (URL)
The URL of a document describing the context of this information.
No machine readable content shall be served at this URL. Reasoning is not
allowed with information from different contexts.
A Audience (URI or blank node)
The intended audience of this information, represented as a class.
Applications shall ignore information if the applications audience (set by the
administrator of the application) cannot be deduced to be a subclass of A.
Arbitrary audiences can be constructed using blank nodes, which are allowed
here.
E Expires (xsd:dateTime)
Applications shall ignore information that has expired. Information that
is considered important or timeless should be "remembered" by changing the
expiration time before it expires (which makes updating the responsible
person necessary).
DNT Do Not Track flag (xsd:bool)
This information shall not be used for marketing purposes if DNT is true.
AD Algorithmic decision policy (enumeration)
nothingToHide: This information may be used for algorithmic decisions
leaveMeAlone: This information shall not be used for algorithmic decisions
about a single person influencing this persons life
youDontKnowShit: Same as leaveMeAlone, but also prohibits algorithmic
decisions influencing the life of an arbitrary group of persons
GS Gaming the system flag (xsd:bool)
If true, this information is intended to game some algorithm and would not
have been generated without it. Applications unprepared for disastrous
feedback loops should ignore this information.
IW Intentionally wrong flag (xsd:bool)
If true, this information is intentionally wrong for the purpose of satire,
irony or otherwise making a point. It can be displayed to humans but shall
not be used for further information processing (reasoning, aggregation, etc.).
IW2 Intentionally wrong intentionally wrong flag (xsd:bool) (submitted by timbl)
If true, the IW flag is intentionally wrong for the purpose of satire,
irony or otherwise making a point. It can be displayed to humans but shall
not be used for further information processing (reasoning, aggregation,
etc.).