ontology-summit
[Top] [All Lists]

Re: [ontology-summit] [ReusableContent] Partitioning the problem

To: Ontology Summit 2014 discussion <ontology-summit@xxxxxxxxxxxxxxxx>
From: Ali SH <asaegyn+out@xxxxxxxxx>
Date: Fri, 31 Jan 2014 12:39:00 -0500
Message-id: <CADr70E1yXtcqoQGJr2=Rz-cXJeJQaqhgsEhXj9mmDdQmirrNrQ@xxxxxxxxxxxxxx>
Dear Simon and all,

On Fri, Jan 31, 2014 at 12:21 PM, Simon Spero <sesuncedu@xxxxxxxxx> wrote:
Using some portion of the URI as a human readable identifier is not ideal.

This seems like an issue that keeps resurfacing year after year. As others have pointed out time and again, and as Tim Finin reminded us with that wonderful link to McDermott's article, human readable names pose the problem that the human interpreter of a URI (or identifier or variable name) may read semantics into the term which are not formally present. Conversely, what David and others have suggested is that, say inspecting a SPARQL query using not-easily-human interpretable names can make the process of writing and validating such queries an arduous (and possibly more error-prone) task.

The question comes down to: can designers of such systems expect the various audiences interacting with aspects of the ontology to ascribe the right semantics to the identifier? Intuitively, it seems that there it should be possible to strike a balance between completely opaque identifiers and potentially misleading natural language-like identifiers. Perhaps some combination of the two?

Where the balance lies is not as clear (to me at least). In my own efforts, I try to combine the two, using elements of NL labels, though with the addition of more opaque elements to emphasize that it is an identifier for a knowledge engineering construct and not a natural language item. I've taken some insipiration from how Wordnet refers to concepts in their synsets, though of course an unscrupulous KE or SME may nonetheless ascribe incorrect / hopeful semantics to the term at hand. It nonetheless reinforces that one should not assign NL-like names without careful thought and some understanding of the intended audiences and what prejudices they may bring to the table.

Is it better to run this risk of hopeful / inaccurate semantics that are read into identifiers, or is better to complicate the task of writing queries / debugging systems by requiring constant referal to an opaque identifier system? If there is a balance, where do you draw the line?

Experiences? Opiniions? Studies?

Best,
Ali

--


(•`'·.¸(`'·.¸(•)¸.·'´)¸.·'´•) .,.,

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2014/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2014  
Community Portal: http://ontolog.cim3.net/wiki/     (01)
<Prev in Thread] Current Thread [Next in Thread>