- From: Bijan Parsia <bparsia@isr.umd.edu>
- Date: Mon, 14 Jul 2003 16:50:54 -0400
- To: www-ws-desc@w3.org
This weekend I tried playing around a bit with modeling services with targetResource and serviceGroup in OWL. Mostly, I wanted to try to concretize some of my intuitions (and, to be frank, my biases :)) Here is the current version: http://www.mindswap.org/~bparsia/ontologies/test/wsdl-target-and- group.owl Some results (not in the formal sense; just on my thinking): 1) It seems to me that targetResource, whatever it is, is more naturally attached to operations, since it is via operations that the Service does anything to anything. 2) But, even then, I don't find targetResource useful. I do think preconditions and effects are (surprise surprise, but it really was for me :)). 3) I think a "logical" targetResource as the "set of things manipulated by the Service or by an operation" tends toward the redundant at best, and tends to miss the interesting things about how services or operations are related. 4) In the absence of other metadata, the concept of a "Service" (qua group of operations) seems uninteresting. targetResource gives the illusion that a Service is a semantically coherent group of operations (in the sense that they all manipulate the same resource), but when the targetResource is just the set of resources manipulated by the operations of the Service...well..I trust the pointlessness is clear. I plan to keep playing with the ontology and will happily try to encode alternatives that people suggest. Oh, while I think preconditions and effects are critical, I'm not suggesting we add them to the WSDL that-might-soon-be-formerly-known-as-1.2. Right now, there isn't a standard semantic web language that most Semantic Web Services folks agree is sufficently expressive for representing the kinds of preconditions and effects one would want to in Web services. DAML-S, for example, basically has punted on this issue, thus far. Cheers, Bijan Parsia.
Received on Monday, 14 July 2003 16:49:59 UTC