David,
Regarding your milk example, it assumes you are going to a store in an English-speaking country. Even then, there is the question of what is the difference between whole milk, skim milk, 1% and 2% fat milk, buttermilk, homogenized, pasteurized etc., not to mention whether we are talking about cow’s milk or goat’s milk or soy “milk” or almond “milk”, or the issue of quantity labels and frames of reference. More importantly, in my opinion, is the fact that you really can’t buy generic “milk”. You can only buy a specific dairy company’s brand of milk, and most big supermarkets offer several brands of milk. That’s important from the perspective of who is accountable if the milk goes bad or has some bacteria or adulteration problem. Note that in most jurisdictions milk cannot be sold without a date stamp applied by the producing dairy. While the consumer might only be concerned about the type of milk and the sell-by date, the actual sales transaction in the supermarket context is for a specific Dairy’s brand of milk. Someone using the milk in a home or restaurant context might view the milk as generic – unless something bad happens. Bottom line: there are a lot of context/scope dimensions and associated values that go into identifying something as simple as a jug of milk, not just the label “milk”.
Similar issues apply to the “postal code” data element and most other data elements one encounters in today’s information systems, like the policy number example. The problem is that people operate on the assumption that the system creating or holding the data element provides those context/scope parameter values implicitly (usually) or explicitly (rarely). But people or other systems usually don’t know a lot of the context assumptions surrounding the systems they are accessing, and are rarely provided any explicit and definitive way for discovering those context/scope assumptions (a cottage industry has formed around techniques for inferencing context assumptions, even though such techniques usually have some non-zero probability of inferencing incorrectly).
So if you encounter something with the word “milk” on its label in a store, do you really know what you are getting if you buy it? Not without a lot more context information. On my morning walk the other day I encountered a half-crushed can of something labeled “Coconut Milk”. Fortunately I haven’t seen such cans in the dairy section of the supermarkets I have visited, sparing me some confusion. J
Hans
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of David Eddy
Sent: Wednesday, September 11, 2013 3:46 PM
To: [ontolog-forum]
Subject: Re: [ontolog-forum] Intelligent ids and postal codes, was ONTOLOG community event planning ...
Ed -
On Sep 11, 2013, at 1:02 PM, Barkmeyer, Edward J wrote:
The problem of 'semantics' in 'postal codes' is the 'intelligent identifier' problem.
Sometimes called "overloading" I believe.
Plus there's the little conflict with the rules of the physical world... when we go to the grocery store, the jug labeled MILK always contains milk.
So far we haven't learned to build information systems like that.
M0760 is a perfectly acceptable computer label & Lord only knows what it means or contains.