Re: [xml-dev] ANN: White Paper - "Using OWL to Avoid Syntactic Rigor Mortis"

[AndrewWatt2000@aol.com]

In a message dated 17/04/2003 16:42:28 GMT Daylight Time, costello@mitre.org writes:


> I have written a white paper[1] titled:
>
>    "Using OWL to Avoid Syntactic Rigor Mortis"
>
> This paper summarizes the discussion on using a logical model (i.e., an
> OWL Ontology) to enable many different physical expressions (i.e., many
> different forms of instance documents).
>
> Comments welcome.  /Roger


Roger,

An interesting, thought-provoking piece. A few comments:

You state,
"An OWL Ontology defines the fundamental relationships of your data.".

With respect this assertion seems to me to be, in certain respects, pretentious, simplistic and potentially misleading.

The term "fundamental relationship" conjures up, in my mind at least, something intrinsic and enduring. In fact the supposed fundamental relationship expressed is an *opinion* or *agreement* by *one group* at *one point in time*. I anticipate the need for versioning of supposed "fundamental relationships".

Such a subjective, participant-sensitive, time-sensitive agreement on usage may have many benefits in terms of expediency or efficiency for certain purposes. And those benefits may, in some circumstances, be major. Yet because of the aspects I mention the process may also fail.

Let me illustrate, using the photography domain you used in your example. You suggested that focal length is synonymous with lens size. Of course that isn't true. An optical engineer would likely complain about that being supposedly synonymous.

There may be practical consequences of inadequately thought through opinions on supposed "fundamental relationships". I have only a few feet from me a 300mm lens which is about 2 or 3 times the size of a 300mm lens I saw in use yesterday. A shipper who made assumptions that my (archaic but producing pleasing optical results) lens is the same size as the latest state-of-the-art lens will either waste a lot of volume in a ship or find that all the lenses won't actually fit into the available space.

So, in my view at least, a "logical design" needs to be specified quite precisely - in terms of its time point / version, who decided it and what aspects of the problem space it is intended to apply to and, possibly more importantly, what aspects it is not intended to apply to.

The issue of stated purpose is also relevant. To take a topical example, a logistics officer in Iraq might be satisfied with a numerical head count of US and UK uniformed staff - it is adequate to tell him how many thousand meals, toilet rolls, and bedding might be needed. A political journalist, historian or Iraqi citizen might find a classification such "liberator" or "occupier" relevant to the aspects of a complex multi-dimensional situation that is of interest to them. There are likely many more purposes for which data in complex, real-world situations might be used. Purpose, I would suggest, is essential to define.

A better restatement of your first definition would be, in my view, "An OWL ontology formalises an opinion of a specified group at a specified time point for one or more specified purposes of perceived relationships among data.".

A nit-pick point. If we are using XML I suggest you  use consistent case on "camera" and "Camera".

You state the first objective of your strategy as being to "maximize diversity of physical expression". Unless I misread your intent what you meant was "allow flexibility in physical expression". As far as I could see there is no intrinsic value in maximising the number of possible physical expressions. Flexibility is the issue, rather than maximising numbers.

Andrew Watt