OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   RE: [xml-dev] Meta-somethingorother (was the semantic web mega-permathre

[ Lists Home | Date Index | Thread Index ]
  • To: "Jonathan Borden" <jonathan@openhealth.org>,"XML Developers List" <xml-dev@lists.xml.org>
  • Subject: RE: [xml-dev] Meta-somethingorother (was the semantic web mega-permathread thing)
  • From: "Michael Rys" <mrys@microsoft.com>
  • Date: Thu, 10 Jun 2004 01:00:16 -0700
  • Thread-index: AcROhuOaE6fsO5NPS1SxuzuvEnlzOwAOU3Pw
  • Thread-topic: [xml-dev] Meta-somethingorother (was the semantic web mega-permathread thing)

Once somebody shows me a description logic inference engine that actually scales, I will become more interested in this aspect.


The one advantage RDF as any extended entity relationship model has over tree models like XML, is the ability to represent relationship graphs. However, for that, there is a very well suited tuple based model: the relational model. Having more semantics associated with the relationships is useful, but only as long as the relationships are built-in.


Best regards

Michael (who in his former live implemented a hybrid database/description logic system)


From: Jonathan Borden [mailto:jonathan@openhealth.org]
Sent: Wednesday, June 09, 2004 6:02 PM
To: XML Developers List
Subject: RE: [xml-dev] Meta-somethingorother (was the semantic web mega-permathread thing)


Begin forwarded message:


From: Jonathan Borden <jonathan@openhealth.org>

Date: June 9, 2004 8:41:34 PM EDT

To: "Dare Obasanjo" <dareo@microsoft.com>

Subject: Re: [xml-dev] Meta-somethingorother (was the semantic web mega-permathread thing)


Dare Obasanjo wrote:


-----Original Message-----

From: Bill de hÓra [mailto:bill.dehora@propylon.com]

Sent: Wednesday, June 09, 2004 2:58 PM

To: XML Developers List

Subject: Re: [xml-dev] Meta-somethingorother (was the

semantic web mega-permathread thing)


I suggest making an informed decison - read the RDF or OWL

Model Theory and decide for yourself whether

XML+Namespaces+HTTP cover it off - if you come back and say

yes, then we'll have something to talk about.


I've read a little RDF & OWL model theory and have failed to see what they buy me above and beyond basing my applications on XML/XPath/XSLT/XSD/XQuery. RDF people like pretending that the XML family of technologies ends with the XML 1.0 spec when this is far from the truth.


One difference, but clearly not the only difference, is that the OWL DL subset allows you to use description logic inferencing engines with "knowledge bases" (i.e. a collection of triples) that conforms to OWL DL. Perhaps this doesn't float your boat, but this is *one* example of something that you can do using OWL that XML/XPath/XSLT/XSD/XQuery doesn't allow you to *directly* do. The reason that I say *directly* is that one can also quite easily come up with an XML format that is OWL DL compatible (e.g. the OWL *XML* presentation syntax).


I suppose for fun you might write a DL inferencing engine in XSLT. Perhaps that could be a new test that Microsoft can give interviewees :-)



I also find it interesting that almost every RDF booster talks about how the benefit of RDF is that you can dump all this wonderful semantic data in an RDF store and then query it.




So where exactly are these RDF stores or standard RDF query languages? I can dump XML data into a relational database today and query it with SQL. I even could dump it in an XML database [or relational database with an XML datatype] and query it with XQuery or XPath.


What exactly do RDF technologies buy me over using XML technologies for doing queries over FOAF + RSS 1.0 data, for instance.


Nothing in that instance. For less trivial examples you can do quite alot in SQL (or XQuery, or even XSLT) with the use of procedures.


The particular value of OWL/RDF is that the model theory licenses a well defined set of inferences given any particular set of triples in the same fashion that SQL defines a set of result tuples based on a set of tables and a given SQL statement.


OWL is great for expressing classifications of things where one is not concerned about whether the "thing" is an attribute or element (in that specific case one can use XSD for example) but where things might be taxonomies of birds, or bacteria or chemotherapy agents, or genes or chemicals or diseases or books or ... anything that you might want to classify. When you make assertions about categories the "system" understand that you are making such assertions about each instance of each of the category's subcategories.


If you aren't that interested in classifying things, then OWL probably isn't that useful to you. (Either that, or you haven't learned about the cool things that you can do when things are classified (assuming that you consider doing stuff like medical diagnosis cool)).



PS: Then there's the fact that RDF doesn't deal that well with mixed content.


There is this fact.





News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS