ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] What is Data? What is a Datum?

To: <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Ed Barkmeyer <edbark@xxxxxxxx>
Date: Tue, 8 Jan 2013 17:11:54 -0500
Message-id: <50EC99AA.1010307@xxxxxxxx>

On 1/8/2013 3:11 PM, Kingsley Idehen wrote:
On 1/8/13 2:23 PM, doug foxvog wrote:
A datum is a role played by a proposition, as John Sowa said.  A single
value is not a datum unless it represents a proposition.
Sure, but doesn't context determine role in this situation? Of course, 
said determination might be implicit rather than explicit which is 
ultimately inevitable. Thus, when _values_ are in the role of 
representing the description of an observation subject, they denotation 
"Data" applies. Likewise, if the aforementioned is achieved via a single 
value, then the denotation "Datum" would apply.

I think this confuses John's assertion that a "datum" is a "role" of a proposition.  He takes the view that the proposition plays a role in the assertion that it is true, and that role is called "axiom" or "fact".  I would just have said that a datum is a proposition that is taken to be, or asserted to be, true.  The context for that role is any context in which the proposition is taken to be true. 

If the context of appearance of the datum is in a report, and you trust the author of the report, then the datum/proposition is taken to be true.  That context has nothing to do with the subject of the report, or the "subject" of the datum, or the the spreadsheet it appears on, or anything the like.  The context is only the acceptance that the proposition is true.

The context of representation of a datum is "quite another thing entirely".  I agree that a "value" may be considered to represent a datum, when the context of its appearance determines both the relation to which it is (in most cases) an argument, and the "subject", which is an/the other argument.  So, in the context of my driver's license, the "value" T labeled date-of-birth is interpreted as the proposition (person.has.date-of-birth EdBarkmeyer T), and that proposition is taken to be true.   (Some "values", such as "true" or "false" and "yes" or "no", are representations of assertions involving unary predicates.  For example, the value "yes" in the blank following "US Citizen?" is taken as a representation of the assertion (person.isUScitizen EdBarkmeyer).) 

So, in such a representation context, each value or tuple of values is in the role of representing one proposition about the subject.  Assuming one respects the observer and takes these propositions to be true, each proposition represents an observation, and "data" is simply a set of observations.  (I think Kingsley's term "description of" means "set of observations about".)  That, unsurprisingly, pretty much coincides with the use of the term "data" in science and statistics.

Propositions (or data objects) are not restricted to triples (subject /
predicate / object,  subject / verb / object, entity  / attribute / value,
entity / key / value, ...).  Some propositions naturally require higher
arity, e.g., X is 3 meters from Y.
Sure, but triples are an effective base upon which higher arity can be 
built, right?

In a word, No.  It is certainly possible to represent a semantically ternary relation as a set of triples, but that is a "form of _expression_".  It is a transformation of the semantic intent into a grammatical structure.  I agree that the idea "semantically ternary" may be dubious, since we are learning more and more about the complex bi-directional relationship between language and thought.  But the point here is that a triples language (we won't name any names) restricts the _expression_ of intent in a way that requires circumlocution for some simple semantic concepts.  That is not at all the same idea as being a "base on which higher arity [of intent] can be built".

Doug said:
Propositions represented by subject -
predicate - direct object - indirect object in English (Juan gave Xue
the book) can be modeled by reifying the action and conjoining multiple
ternary propositions, but a single higher-arity proposition can be useful
for many purposes.

As I said above, it is more than that.  An atomic idea, like A being between B and C, does not conceptually involve multiple relations.  Of course, it is always possible to do the Davidsonian thing:  There is a between-ness S, and the agent of S is A and the patient of S is an ordered pair P and the first member of P is B and the second member of P is C.  But then that pattern can also be applied to "Cain killed Abel", which a triples language can, and does, represent atomically.  The problem is that the triples language does not formally adopt a Davidsonian representation of states with a fixed set of binary relationships.  It simply adopts circumlocutions when it cannot represent an atomic concept atomically.

The single higher-arity proposition is useful because it is semantically atomic.

Sure, but you always need a building point that provides foundation (be 
permanent or temporary).

Well, you need a grammar for every representation language.  That is the structural foundation.  If you choose a grammar in which verbs have only a subject and a direct object, you will force a lot of thoughts to be phrased using circumlocutions, because there is nothing natural about that restriction.  Indo-European languages have evolved a number of other grammatical elements to augment binary verbs in expressing intent.  Triples languages have not.

-Ed

-- 
Edward J. Barkmeyer                        Email: edbark@xxxxxxxx
National Institute of Standards & Technology
Systems Integration Division, Engineering Laboratory
100 Bureau Drive, Stop 8263                Tel: +1 301-975-3528
Gaithersburg, MD 20899-8263                Cel: +1 240-672-5800

"The opinions expressed above do not reflect consensus of NIST, 
 and have not been reviewed by any Government authority."



Kingsley
-- doug

On Tue, January 8, 2013 08:13, John F Sowa wrote:
...  The word 'proposition', for example,
is more fundamental than the words 'assertion', 'statement', 'judgment',
'assumption', 'belief', 'hypothesis', 'axiom', or 'theorem'.  Therefore,
it is reasonable to say that Proposition is the natural type, and the
other words describe roles that a proposition can play.
The words 'datum' and 'data' most definitely describe roles.
On Mon, January 7, 2013 19:29, Melvin Carvalho wrote:
On 7 January 2013 23:25, Kingsley Idehen <kidehen@xxxxxxxxxxxxxx> wrote:
  On 1/7/13 5:04 PM, Gary Berg-Cross wrote:
...
"data object" is much clearer [1] i.e., a resource comprised of
structured
data. Typical representation, for a given perception medium (e.g., the
World Wide Web or paper) is a subject->predicate->object,
subject->verb->object, entity->attribute->value style of graph pictorial
:-)
I used to talk to people about "predicate / object" and mainly would get
blank stares.
I tried "attribute / value" which seemed to have *slightly* more
understanding.
I'm adding to my terminology to term "key / value pairs" which I think may
be effective to some audiences

        
Links:

1. http://bit.ly/PnTJdV -- understanding data objects .

Kingsley


Gary Berg-Cross, Ph.D.
gbergcross@xxxxxxxxx
http://ontolog.cim3.net/cgi-bin/wiki.pl?GaryBergCross
Potomac, MD
240-426-0770
  
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
  





_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>