IBM and University of Georgia have recently submitted a notes to W3C for the description of the semantics of a Web service.
The title of the note, “Web Service Semantics – WSDL-S”, is somewhat misleading, but the idea proposed really worths some attention.

That’s pretty much all you can tell using native WSDL. The consumer of the PO services now knows it need to send a message to you as *syntactically* defined by the GED “tns:processPurchaseOrderRequest”, and it will get something back as defined by GED “processPurchaseOrderResponse”.

           <!Category is added as an extensible element of an interface>
         <!Precondition and effect are added as extensible elements on an operation>
There is indeed one glitch, but it’s easily fixable and I won’t worry about it: the proposal is based on an old version of WSDL2.0. It needs to be updated to reflect the latest WSDL 2.0 draft. . It’s worth pointing out that it also supports WSDL1.1.

Related standard initiatives

The proposals for creating two new W3C working groups are being circulated in W3C membership. One for Web services semantics characterization which is to define requirements and scenarios, and the other to standardize WSDL-S.

Don’t confuse WSDL-S with some other initiatives
    • OWL-S. It provides a language for actually defining the semantics of a services. WSDL-S may point to an instance of OWL-S
    • WSDL2.0 RDF mapping. It provides an OWL ontology for WSDL2.0.

To report this post you need to login first.


You must be Logged on to comment or reply to a post.

  1. Bernd Eckenfels
    Kevin do you think those annotations  are needed? I mean one can easyly use the RDF approach where you point from the outside into the type definition “I want you to explain that the element ns:X is a Y”, so no extensions to the WSDL is needed. After all, thats what RDF is all about. It might be good to be able to find those semantic descriptions which are related with a WSDL, but that could also be done in the registry.


    1. Kevin Liu Post author
      Hi Bernd,

      A very valid question. Nobody wants yet another way to do things when there is already a good solution. In the case of WSDL-S, I think it is a nice addition to WSDL which is well adopted in the Web services community, and will help bring the semantics aspect into the picture.

      You are right that one can use RDF to point to a WSDL, or use a registry to connect a WSDL and a semantic description, or use some other mechanisms. However, the processing model for each of the approach would be quite different. For example, pointing to a WSDL from RDF and pointing to a RDF from a WSDL may be considered as the same thing from the functional point of view, but the processing may be very different. At least from tool interoperability piont of view, a well accepted simple mechanism would be very helpful. Built-upon the popularity of WSDL, WSDL-S seems to have good potential.

      A analogy may explain a bit. If you are familiar with the WS-Policy work, you know that it defines a WSDL attachment mechanism. I consider WSDL-S as a sort of WSDL attachment mechnism for We service semantics.

  2. Anonymous

    Describing WS semantics surely is beneficial for the  expressiveness of the interface. Even just the annotation of inputs and outputs is a great advance. I wonder if WSDL-S provides something to overcome structural differences between a WS schema and a semantic schema. It is highly unlikely to find a corresponding concepts in the semantic spec that exactly matches the input/outpt of the service. I think this is one of the key factors for every notation. What is your opinion?



Leave a Reply