SDWP 2005 Panel Discussion

Monday, July 11, 2005 -- 1:25-2:10 p.m.

 

Alternative Frameworks for Semantics in Web Services:

Possibilities for Inclusion in Future Standards

 

Moderator

 

Panelists

  1. Richard Goodwin
    IBM T. J. Watson Research Center
    rgoodwin+us-ibm-com

    Statement: Semantic web services will only succeed if they disappear. People creating service oriented applications simply want to deploy applications quickly and have them to work reliably. Semantic meta-data can support these goals, but won't be used if it imposes an additional burden on practitioners. The semantic meta-data needs to be invisibly captured as part of the service design and creation process and not thrown away as is currently the case. By retaining the semantics from system designs, UML models and other artifacts, we can provide semantically rich service descriptions and integrate them into the current web service infrastructure.

  2. Alain Leger
    France Telecom R&D
    alain-leger+rd-francetelecom-com

    Statement: Semantic web services present a very promising avenue for a large spectrum of applications with a clear ROI and many benefits for everyday life activities. But is the global concept new? Certainly it is not (e.g. Distributed computing, DAI, MAS, ...). What was the adoption in the industry? So, we believe that a rapid take-up in industry should be a major concern within the SWS community. To that end: showing the business value on concrete business needs and use cases, focusing the research effort on still key roadblocks, hiding the complexity of the technology, making available automated tools and standard compliant frameworks, and standardizing on key SWS elements should pave the way towards fast adoption in industry.

  3. Ling Liu
    Georgia Institute of Technology
    lingliu+cc-gatech-edu

    Statement: Semantics in web services can play a critical role in the rapid deployment of service oriented applications. However, this will happen only if the use of semantics will not add extra burden on the developers and users of the web services, and yet demonstrate real value in terms of productivity. It is widely recognized that relational database systems are successful in introducing relational data semantics into the database world, which facilitate the physical and logical database design, and demonstrate the increased productivity of both database developers and users. We will use an analogy with relational data semantics in applying semantics to web services. We argue that there is a need for a location-independent service-oriented execution model that facilitates the rapid design and deployment of service-oriented applications, enabling the effective creation, execution, and composition, as well as automated discovery and classification, of web services.

  4. Michael Stollberg
    DERI Innsbruck
    michael-stollberg+deri-org

    Statement: Web Services promise a novel paradigm for service-oriented architectures over the Web: Web Services shall provide computational facilities accessible over the Web that are automatically and dynamically combined and utilized for specific application scenarios. The current Web Service technology stack around SOAP, WSDL, and UDDI fail in realization of this promise: Web Service discovery, composition, and usage require manual inspection and integration. Semantic Web Services aim at overcoming these deficiencies: using ontologies as the underlying data model for ensuring semantic information processing and interoperability, inference-based mechanisms are envisioned for automated discovery, composition, conversation, and execution of Web Services. The WSMO/WSML/WSMX approach presents an complete framework for Semantic Web Services, including semantic descriptions of core elements along with a language for this, technologies for the main Semantic Web Service functionalities, and an integrated execution environment.

  5. Andreas Wombacher
    University of Twente
    a-wombacher+utwente-nl

    Statement: Web Services based on SOAP, WSDL and UDDI are not sufficient to handle more complex and state dependent services. In particular, the service discovery has to take further aspects into account like for example the semantic description of the provided functionality, the description of the behavior of the service, or Quality of Service guarantees provided by the service. However, extending service discovery by these descriptions has to take the following issues into consideration: (i) to be able to use these descriptions somebody has to provide them, which mostly requires at least partly manual work which costs a lot of money; and (ii) to keep the resulting service discovery usable it must provide fast response times and accurate results. Thus, the computational complexity introduced by handling the descriptions and the correctness of the results has to be considered. Having these issues in mind I expect semantic approaches supporting elaborated reasoning capabilities to have too high computational complexity and semantic approaches with less reasoning support to have too imprecise results.

 

Issues

  1. The OWL-S Approach

  2. The WSMO / WSML / WSMX Approach

  3. The METEOR-S / WSDL-S Approach

  4. Other Approaches (e.g., MDA/UML)

  5. Evaluation: Compatibility, Conciseness, Complexity, Practicality, Useability

  6. Proof of Concept: Prototype using Realistic Examples

  7. Pathway to Industrial Acceptance

 

[2004 panel ]