|
On 7/25/13 11:14 AM, David Eddy wrote:
Kingsley -
On Jul 25, 2013, at 10:38 AM, Kingsley Idehen wrote:
What in the SW allows me to know, enables me to
learn that the labels "M0760" and "MENSA-FL" mean
the same thing?
Yes, via an owl:sameAs relation that associates
relationship (triple) participants with the labels
"M0760" and "MENSA-FL" .
Here is an example that leverages Linked Data i.e., the
use of HTTP URIs to denote the relationship participants
such that a human or machine has access to description
oriented data that extends from relationship
participants all the way to the actual relation
semantics:
## Turtle ##
## Co-Reference example ##
<#Thing1>
a <#Entity> ;
<#Thing1> <#label> "M0760" .
<#Thing2>
a <#Entity> ;
<#Thing1> <#label> "MENSA-FL" .
To my eye, you are stating a semantic equivalence... who or
what does the discovery?
I apologize for being unclear. Too much ambiguity in the
air.
It is one thing to have a mechanism that STATEs equivalence
or similarity across thingies that is obviously easily solved
in a variety of ways. A spreadsheet would suffice &
likely be far more familiar to most folks.
I am demonstrating how (using Semantic Web stack components) you
deal with co-reference. You have a relationship comprised of two
entities facilitated by an owl:sameAs predicate. HTTP URIs denote
the object, subject, and predicate of the triple that represents the
relationship.
The challenge is how do I LEARN or DISCOVER that thingies
are same/similar/related. How does SW do semantic DISCOVERY?
In one of my live examples (see the G+ note links) I have a live
demonstration of owl:sameAs inference. Basically, the reconciliation
is automatic.
If for whatever reasons you don't like the reconciliation you can
make your own set of statements and then use them as the inference
basis.
Currently that sort of esoteric information is typically
walking around in the heads of SMEs who can be very difficult
to pin down.
We need to find a way to settle them down and grab their attention.
The problem with attention, in this day an age, is that its really
hard to capture. People have too much coming at them which is why
(as John Sowa articulates, repeatedly) these tools have to build on
the legacy infrastructure in use, that's the only way to grab
attention.
The situation is that I'm new to "the system" (a highly
ambiguous & imprecise description since I really do not
know enough to understand if I'm dealing with A system or a
dozen systems) & need to achieve some sort of
interoperability of various pieces of data. I want to pull
data out of A, B, & G & merge it into X. Very typical
activity.
Better tools.
We have built a lot of these tools. I've written a lot of demos. I
just need others to join the few of us that take the dog-fooding
approach with end-user, integrator, domain expert etc. tools :-)
Assuming I am working in a typical 20-30-40+ year old
(software/documentation) collection, can I just point
owl:sameAs functionality at 50M LoC (lines of code in several
dozen languages) & it will tell me, or ask me to consider
that M0760 & MENSA-FL are potentially the same? I'm
guessing SW is silent on this challenge.
No, you would transform the legacy data (using transient or
materialized views) and then apply the reasoning at the view layer.
What you cannot do is "rip and replace" their existing systems. This
is where the problem lies re. "the Semantic Web technology stack"
since many advocates overtly seek "rip and replace" over "embrace
and extend" .
Kingsley
_________________________________________________________________
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
--
Regards,
Kingsley Idehen
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen
|
smime.p7s
Description: S/MIME Cryptographic Signature
_________________________________________________________________
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)
|