- From: Sankar Virdhagriswaran <sv@hunchuen.crystaliz.com>
- Date: Fri, 28 Apr 2000 11:10:09 -0400
- To: <www-rdf-interest@w3.org>
Folks, I am sending an abstract of a document that we are working on to pique your interest. I would be interested to hear from folks who would like to work with us on this. We have a very, very preliminary draft of the document. But, what we do have is a query engine that leverages the ORDL language to support exploration of structured information. Interested? Let me know. ----------- Object Relational Description Language (ORDL) Abstract Object Relational Description Language (ORDL) addresses the need to describe and represent information objects and relationships between them such that automated agents or users could perform guided exploration over the information described by ORDL. ORDL is used by content providers to explicitly describe and represent information objects and relationships of various kinds starting from primitive types all the way up to complex, networked structures. We expect ORDL to help enable a semantic web that supports more intuitive exploration and evaluation. Background and Motivation The need to describe information objects and relationships between them is well recognized. The field of knowledge representation, modeling languages for databases, and ontology representation efforts has tried to address this need. Resource Description Framework (RDF) is one response to addressing a part of this need. While RDF focuses on describing relationships between resources, ORDL focuses on describing information objects and representing relationships between them. While RDF is relationship centric, ORDL is information object centric. Furthermore, while RDF uses a relational model that allows world wide scalability by allowing anonymous definitions of relationships over information resources, ORDL is targeted more for describing objects and relationships by parties that control (directly or indirectly) information objects that they are trying to relate. While RDF is ideally suited for describing information about web sites or applications that one does not own or control - i.e., not under ones administrative control, ORDL is ideally suited for describing relationships between information objects that is under one's administrative control. RDF's data model starts from relationships (tuples) on top of which type and other constraints are added, while ORDL data model starts from a traditional object oriented data model and relaxes it to deal with semi-structured nature of information found on the Web. Furthermore, while RDF is about meta-data, ORDL is about describing meta-data and representing instances of data that follow the specification of the meta-data. Because of this, ORDL uses XML-Schema as its serialization representation. ORDL Data Model Introduction Features, Categories, Types, and Entries The ORDL data consists of typed features that are loosely organized into a heterarchy of categories. Features may belong to multiple categories and features may exist without ever being part of a category. A category can optionally be designated as a child of one or more other categories. Child categories inherit features associated with their ancestor categories. Additionally, ORDL uses a reflective model. Hence, its schema (i.e., meta data) descriptions can be extended the same way instances are extended.
Received on Saturday, 29 April 2000 18:35:07 UTC