| [ Team LiB ] |
|
12.2 Brief History of the Ontology MovementThe current ontology effort at the W3C is OWL, which was rooted in the DAML (DARPA Agent Markup Language) project — specifically the ontology language originating from this project: DAML+OIL. Though you'll see DAML+OIL, OIL (Ontology Inference Layer) originated separately in Europe starting in 1997. It was preceded by SHOE (Simple HTML Ontology Extensions) in 1995. OIL is particularly relevant in a book on RDF because it was the first ontology based on RDF as well as the XML Schema. The first release of DARPA's DAML was in 2000, following early standardization work on the part of the W3C. One year after DAML was released, a joint ontology language, DAML+OIL, was released. An early press release on DAML+OIL appearing in the Cover Pages (at http://xml.coverpages.org/ni2001-03-28-a.html), said the following:
The first version of DAML+OIL was released in December of 2000, and the current version was released in March 2001. At the time, one of the primitives that DAML+OIL provided was data typing, which the first RDF specification didn't provide. What's interesting with DAML+OIL is that the classes and properties and their relation to each other as defined in the document are extremely similar to those shown in RDFS (as described in Chapter 5). For instance, a daml:Class element categorizes elements that are classes. There is also the concept of property, defined through daml:ObjectProperty, but there is a conceptual difference between class and property in DAML+OIL and class and property in RDFS. However, exactly what this conceptual difference is has been the focus of considerable debate within the Semantic Web community. In the www-rdf-logic mailing list, a thread started once about the difference between rdfs:Class and daml:Class (at http://lists.w3.org/Archives/Public/www-rdf-logic/2002Mar/0017.html). Exactly when does one use rdfs:Class and when does one use daml:Class? General consensus tends to support the view that RDFS describes metadata, including DAML+OIL itself. However, one should use DAML+OIL elements to define actual instances of data, such as elements as they are defined in my PostCon vocabulary. This does make sense and supports the view that I have of DAML+OIL as compared to RDFS: that DAML+OIL is a way of describing a generalized business model, such as those defined in PeopleSoft and SAP, while RDFS is the metalanguage that defines DAML+OIL, equivalent to the relational data model used to define the databases that support PeopleSoft and SAP. Unfortunately, this view hasn't received complete concurrence from all parties, and there is considerable bleed-through of the use of one schema over the other or, said another way, lack of clarity between the layers of the architecture, using the parlance of the community that works closely with ontologies and RDF. The W3C entered the picture more fully when it formed the Semantic Web Activity Group in February 2001 and followed up with the creation of the Ontology Working Group. An announcement in August 2001 revealed the intent of incorporating the work of the DAML and OIL groups into the W3C Semantic Web activities (from http://lists.w3.org/Archives/Public/www-rdf-logic/2001Aug/0014.html):
Though not necessarily a part of the effort, DAML+OIL provided the foundation for the W3C ontology effort. The Web Ontology (WebOnt) Working Group was formally launched in November 2001, and work began on defining the language necessary for an ontology layer—OWL, the Web Ontology Language.
 |
| [ Team LiB ] |
|
