Global Grid Forum, Edinburgh, July 2002
David De Roure (U Southampton),
Carole Goble (U Manchester)
The Semantic Grid (SG) BoF was introduced by DDR - outlining the main reasons why this was important to take existing Grid efforts (generally aimed at Computational Grids) further. DDR suggested that there was now a need to bridge Semantic Web efforts and the Grid communities - and to inquire whether these are two entirely different "futures" (or related to each other)
DDR and CG suggested that "XML had become the new ASCII" - and illustrated this from a W3C diagram. CG outlined the need for supporting metadata - primarily data intended for machine consumption. The importance of an "ontology" to define semantics was also outlined - primarily to allow resources to interact. CG suggested that supporting an inference engine with the ontology was important.
CG suggested that one of the objectives of Semantic Web efforts were to investigate standard ways of defining metadata and supporting reasoning - and the understanding of how these could be taken forward in the context of Grids was important.
Comments from the audience:
Where do we draw the borderline - in terms of describing the service. How the inference is to be supported in the context of services (state) - may be too complex.
DDR/CG suggested that the group was to explore issues associated with Grid services - as related to themes being addressed in the Web Services community.
Comment:
Workflow example - time to execute complete workflow (estimate?).
When to do inferencing and when not to? (Is a simple vocabulary another instance of ontology). Does one need to define a taxonomy/classification of services?.
CG gave the example of the most common ontologies used today - Yahoo categories.
CG also suggested that we should not simply explore Grid services as being offered by software, but also as expertise offered by people.
CG/DDR suggested that good vocabularies in a particular application domain were important, particular cluster of people who will agree to decide a vocabulary.
Comment: At present there is no one Semantic Web but multiple specific Semantic Webs (enterprise Grids?) - should this also hold for Grid Computing?
Comment: Should the group explore emergent semantics in the context of the Grid community - not the larger Semantic Web community.
CG suggested that the group would particularly explore the cases where the metadata or ontology breaks down. An example sites was the 404 example from HTTP -- .Metadata breaks. Ontologies break. Ontology 404.
DDR mentioned about the existing work in this area - initiated through participants from a panel at WWW 2002 (May), Ian Foster, Eric Miller, Jim Hendler, Carole Goble.
W3C (Web-Ontology)_group, part of the W3C standardisation effort - result of the DAML + OIL efforts http://www.ontoweb.org/ (resource about ontologies, survey of ontology tools, ontology evaluations)
Earthquake Grid project (workshop in November). Identify common themes that could help build a community ontology. This group will not focus on working towards domain specific ontologies.
Incremental rollout of tools and technologies:
Research Group or Working Group?
It was decided that the group would aim to deliver a Primer on Semantic Web for the Grid community - as an initial document. Will cover what really exists - and what does not.
Exploration with selective other working group on what are the common themes:
Exemplars taken to the Applications WG (to identify and what their comments would be).
DDR and CG also indicated that there was no .stylised. form on interactions between different working and research groups at the GGF at present. Perhaps this should be looked into. Outreach into other groups.
Objectives (milestones) that are short term (6 months down the line). The objective of developing an initial primer. Existing work is not amenable to the Grid community.