On 2/9/09 9:16 AM, Juan Sequeda wrote:
Of course... but even though. IMO, not easy enough! I'm taking the
position as owner of one of the million web applications out there,
powered by a rdbms, and now hearing about the LD thing going on. If I
want to be part of it... I would have to invest a lot of time and
effort with existing tools such as d2r sever, etc...
Juan,
Your views are quite "subjective". Nothing wrong with being subjective,
but it's not easy to digest when they are presented as the norm.
In 1992 NeXT had the ability to produce a "Domain Ontology" and instance
data for such ontologies from RDBMS engines via Enterprise Object
Frameworks (EOF). The only problem was that the solution wasn't DBMS
independent (they supported Oracle and Sybase) and the output was
landlocked in Objective-C land. Also remember, the entire Web originated
on a NeXT machine. I nearly got NeXT to use our UDBC (precursor of all
ODBC implementation outside Windows) as part of an effort to open up
this solution but Apple bought NeXT before we could get it all sorted,
and once part of Apple, the whole EOF project started a steady death
march to what is now "Core Data Services" (a desktop oriented solution
based on a graph model atop SQLite).
The sooner we understand where we are coming from, the easier it will be
to chart where we are headed. If we want to reinvent the universe using
a couple of new labels then good luck to everyone. Expect to ask the
same questions based on the same frustrations this time next year, and
every year after.
I am a firm believer in reuse of knowledge. At this juncture the
problems lie in the general tendency to say: I don't like that, it's too
hard, I am going to build or look for something simpler. Sadly, the net
effect is a form of NIH that leads to: attempting to rebuild the
universe and all the complexity inherent in the process.
If you feel the current solutions are hard, what about attempting to
write a layer atop any of the solutions that work best for you? The net
result will at least benefit some community of users. Simply taking the:
"its too hard" point of view, in a very generic but inherently
subjective way simply doesn't solve anything whatsoever.
The RDBMS to RDF technologies need to be able to do the following
(automagically):
1. Produce basic "Data Source" ontologies for RDBMS Schemas (*press
button post selection of RDBMS schema objects*)
2. Generate Instance Data for the generated "Data Source" ontologies
(*which creates an RDF View over the RDBMS layer*)
3. Publish RDF Linked Data for the resulting RDF Views
Don't mean to be harsh, but I traveled a long way re. these matters, and
it does ruffle my feathers when I see the same patterns and mistakes
playing out. Remember, these issues aren't new, and researching them
didn't start in 1998.
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President& CEO
OpenLink Software Web: http://www.openlinksw.com
