[Top] [All Lists]

Re: [ontolog-forum] Relating and Reconciling Ontologies

To: "[ontolog-forum] " <ontolog-forum@xxxxxxxxxxxxxxxx>
From: David Eddy <deddy@xxxxxxxxxxxxx>
Date: Fri, 22 Apr 2011 14:47:22 -0400
Message-id: <D32A38E7-5AD8-4655-9252-A56D0BE1A01A@xxxxxxxxxxxxx>
Jack -    (01)

On 2011-04-21, at 5:15 PM, Jack Park wrote:    (02)

> Now, you say there are actually some 70 different "labels" (aka name
> strings) for that concept, one of which is "M0101.."  (which makes no
> sense to me as a label for a key: sounds more like one of the values
> associated with that key).    (03)

I certainly did not intend to imply M0101 is a key other than to some  
magical access mechanism into your brain as to what it means.  It  
would be no more a database key than M9999.    (04)

Perfectly logical my Dear Watson.    (05)

When you're constrained by Fortran's 6 character names, M0101 is just  
M = masterfile
01 = first segment
02 = first field    (06)

The masterfile description was 64 pages long... about  1700 - 1800  
data elements.    (07)

The EXACT duplicate of M0101 was MSTR-POL-NO... which I assume is  
perfectly obvious... well, to a COBOL programmer, maybe no so much to  
a CEO or CFO.    (08)

I dearly wish I'd been prescient enough to get a list of the other  
names just to see what else they had.    (09)

The reason for such diversity is essentially the same as today... the  
company had built systems, bought systems, bought systems & modified  
them, bought other firms, etc.  Normal dynamics.  Keeping the  
important data names in sync was not then, is not now, & never will  
be a CEOs priority.    (010)

But there is a business issue here that the plethora of names  
definitely hurts (read: increases operational & understanding  
costs).  How do you go across all those silos with different labels  
for the same thing when you need to know if someone has bought more  
life insurance from you than you wish to insure yourselves?    (011)

> Ok. Le't assume we are just talking about that key. Lets put it into a
> frame-like representation to see if I got it right.
>     Frame: rdf:ID = "<some uri>"
>           commonLabel: policy number
>           otherLabels: M01010, .....
>           domain: ....
>           range: .....    (012)

Assuming you'd have something like    otherLabels: M01010, MSTR-POL- 
NO, contract_ID, ....   ?   When I search for one I find all?    (013)

I'm not sure if I'd go with "commonLabel"... there are HUGE issues  
here in how & what people remember & use for language.    (014)

I clearly remember attempting to discuss the finer points of MSTR- 
MENSA-FL (don't ask... it was 6 months of my life) with the SME.  He  
clearly had no idea what I was talking about.  Then he picked up some  
clue... "Oh... you mean M0760... now I know what you're talking  
about."   (My thought... how does one know if M0761 is or is not the  
same as or different from M0923?)    (015)

Where is that ontological app that I can have on my iPad that will  
automagically sync with the 30 year SME to tell me he's thinking  
M0760 and let him know my MSTR-MENSA-FL is actually his M0760?    (016)

This is where I loose how ontologies provide value... at least to the  
care & feeding of legacy systems (the systems that run our lives).   
What I hear here is an attempt to find the perfect, correct concept  
(name/label in my domain).  There is no such thing.    (017)

For 20+ years I've watched folks tilt at the "one correct standard  
name across the enterprise".   Never going to happen.    (018)

I posit that we need something that discovers the names/labels in the  
context of their native use (my SME having worked with M0760 for 30+  
years) & associate them with the other similar/like things.    (019)

So far I have not seen any interest in the ontology world for this  
issue.  Can ontologies be relevant to dealing with legacy systems?    (020)

David Eddy
deddy@xxxxxxxxxxxxx    (021)

781-455-0949    (022)

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    (023)

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