On 2/9/09 10:39 AM, Andreas Langegger wrote:
I'm afraid you cannot do this "in an automatic fashion" without
knowing what to expose as RDF actually.
You can make the "Data Source" ontology automatically, but nothing
beyond that (the point you are making).
We made the strategic error of dealing with the reality of complex
mappings thereby skipping the vital initial step of automatic "Data
Source" ontology and instance data generation. The incoherence about our
SPASQL based Meta Schema Language is the net result :-(
Replies to the complex (or beyond "Data Source" ontology mapping) below:
For me there are three poeple with real clear incentives:
1. news corps and those bigger publishers who already have their
developers or can pay some to do that mapping (BBC)
Yes, but even in these cases, making a "Data Source" ontology from the
RDBMS schemas driving existing content management systems can be
helpful, even if it simply proves:
1. It can be done
2. It can be done in a manner that scales (*this is the heart and soul
of the matter, ultimately*).
2. research orgs, NPOs, NGOs, also GOs which publish public data
(EuroStat, US census)
Ditto.
3. nerds like us ;-)
Hmm. for the nerds, this is where the problem lies. Most nerds of type:
programmer, don't think data modeling first. They think in terms of
languages and programming constructs first :-( Thus, what is basic stuff
of a DBMS style architect is complex to the programmer. The "nerd"
wants a tool that simplifies life in his/her universe-of-discourse, but
fatally assumes that his/her subjective world view is the norm :-(
So when all is said and done: Context is King!
But any other people like bloggers, etc. will have to wait until the
nerds have created CMS with RDF built-in (e.g. Drupal), but they won't
be able to expose data they already have in RDBMS without us nerds.
Unfortunately not true. Do you have an idea of the size of the following
realms:
1. Informix 4GL developers
2. Progress 4GL developers
3. Sybase / PowerBuilder developers
4. Visual Basic atop ODBC developers
5. Oracle Forms developers
6. Ingres 4GL developers
7. etc..
The communities above grok the following:
1. DBMS technology and best practices (how do describe data containers
as well as navigation via cursors of different types)
2. Security and Scalability
3. Realm specific Entity-Attribute-Value + Classes & Relationships atop
RDBMS substrates (but in application specific ways)
4. RowIDs (URIs in Linked Data land) and Keys (Unique or Primary) as
Indirect Identifiers (IFPs in Semantic Web and Linked Data land)
5. Identity
Drupal is a baby tool the developers from the communities above. Ditto
many of the LAMP application stack.
The real issue is getting the communities above to realize that Linked
Data solves a problem they already know pretty well. All we need to do
is connect terminology and show how we've shrunk the "Report Writer"
courtesy of HTTP and its ability to deliver negotiable data
representation (presentation).
Kingsley
regards,
Andy
On Feb 9, 2009, at 4:30 PM, Juan Sequeda wrote:
On Mon, Feb 9, 2009 at 9:20 AM, Kingsley Idehen
<[email protected] <mailto:[email protected]>> wrote:
On 2/9/09 10:12 AM, Juan Sequeda wrote:
Do you know any solutions that produce RDF Linked Data from
RDBMS sources? Any that exist right now and can deliver such
capability across any ODBC or JDBC accessible DBMS? What
about such a thing even going as far as providing working
examples for every major DBMS engine?
Not yet... unfortunately. There are still open research
problems: ontology matching, linking algorithms. All of this is
needed to have such a system.
At what time in the future do you expect such a product to
emerge?
When such a system exist, is when we can say that everybody is
on the semantic web. Time... how about by the time I finish my
PhD. :P
Well you've just made my point re. Cognitive Dissonance :-(
I don't agree. I am clearly stating a problem, and the paths needed
for it to be solved. How am I contradicting myself?
Ok.. I believe I see where the confusion is. Of course virtuoso and
d2r server publish RDF linked data, etc. What I mean is that it does
not happen in an automatic fashion. That is needed! At least for all
my web developer colleagues who will never invest time in learning
about virtuoso and d2r server.
Kingsley
--
Regards,
Kingsley Idehen Weblog:http://www.openlinksw.com/blog/~kidehen
<http://www.openlinksw.com/blog/%7Ekidehen>
President& CEO
OpenLink Software Web:http://www.openlinksw.com
http://www.langegger.at
----------------------------------------------------------------------
Dipl.-Ing.(FH) Andreas Langegger
Institute for Applied Knowledge Processing
Johannes Kepler University Linz
A-4040 Linz, Altenberger Straße 69
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President& CEO
OpenLink Software Web: http://www.openlinksw.com
