Semantic Grid BoF – 16th October 2002 Chicago
Carole Goble (U Manchester) Chair
Dave De Roure (U Southampton) Chair
Minute takers: John Brooke (U of Manchester), Steven Newhouse (Imperial College)
Present: Who is present (23 people), and briefly why they are interested (impossible to capture all the information for this)
HP Life Sciences
C Harry – Novell
Bill Johnson – SFSG
?? Web
Markus Hart –
Simone ?? - Brunell
Anskai
Ohio Statue University – Bioinformatics
Darcey Quensel – CANRIE
Jon MacLaren – U Manchester: Where is semantics coming from
Paul Watson – U Newcastle : Computation Data
Global Biodiversity Facility Electronic Catalogue
Wolfgang Ziegler – Sci comuting & algorithms. Grid Dictionary
Steve Newhouse (Imperial College) ontology of services and resources.
John Brooke: UoM Service Discovery to promote interoperability.
Jill Coffman : IBM Life Sciences
Geoffry Fox
Dave and Carole outlined the vision of the Semantic Grid along the lines presented at Edinburgh (GGF5). This is the second SG BoF.
Question from floor: is metadata of Semantic Web for humans or machines? CG answered it is for machine consumption but respects semantic meanings recognised by humans.
2nd question, what about differing definitions of, say, hurricane in climate research. CG answers that this is what SG is about. Contrast lexical translation with equivalence by semantic meaning. Use Semantic Web technology to reason about services. See semantic web services (DAML-S)
E. Ganguly from Computer Associates asked for an example of an ontology, CG directed to web site for more information.
Question from floor: can ontology be used as a full knowledge model, answer yes.
E Ganguly asked what was specific about a Grid Service rather than a Web service, answer from CG was that there were Grid concepts such as state that went beyond vanilla Web services. John Brooke noted that therefore ontology had to include information provided dynamically by functions e.g. by some query mechanism. Semantic Web is related to static data description. Semantic Grid is concerned with dynamic services
Point was made very strongly by CG that it is about information integration not AI, though AI techniques can be used. In particular techniques from the database community are highly relevant. It is about information integration. This is a tractable problem. AI is not.
CG proposed that aim of RG is “using Semantic Web technologies to profit the Grid (rather than acting as a driver for Semantic Web”.
Question, is it about aligningWeb Services with Semantic Web, CG answered that there is already much research on this. Steve Newhouse described XML description of port in Grid service to provide information about service. Currently OGSI states that service description is via name, but CG pointed out that this is not a basis for an ontology. Example of Condor resource selection via imprecise matching.
Question: Have there been discussions of structure of ontology language. Are web services associated with the semantic web? Not necessarily… but Grid Services provide a mechanism for provide information to encapsulate an ontology describing service data and to describe the service semantics.
Meta-data are noun like and service interface are verb like. Limited derivation from one ontology to another
DDR pointed to the documents available on Semantic Grid website, new journal from Elsevier.
Steve Newhouse (Imperial College): current Charter does not mention much about relation to Grid. Paul Watson pointed out that it read one way SW->Grid. PW where would the emphasis on services come, CG replied that Web Services was now driver for Semantic Web.
Proposal is to use a sentence after “researchers” and before “This RG..” to get more Grid into it.
Steve Newhouse, will there be any documents, answer: this comes in milestones.
Jon MacLaren (U of Manchester) noted that Point 4 should say “encourage SW community to be involved in SG effort..”
Bill Johnston (LBL, IPG, co-Chair of Architecture Area) should use other WGs to test and ground concepts developed in SG-RG. Offer relevant SG support e.g. ontology experience to other groups. Pushing into other WGs , pulling scenarios and case studies out or other types of defined tasks. Could produce lots of workshops and some document activity. For example on Point 6 of the charter. As an example of possible links, the scheduling dictionary is moving towards an ontology. RG has a service nature to other WG/RG’s e.g. Scheduling Dictionary-WG. Use push/pull model between different groups. Missing a ‘vision’ as to what the semantic grid is – what is gained from this RG. Need a definition of the SG? Is the RG covering Data & Services. Clarification/expansion needed.
Bill, needs expression of vision of what SG does for end user. An illustrative example of benefits eg ontology applied to given services.
Steve Newhouse: clarify what ontologies looking at, is it data and services? If so make it clear maybe via examples as in previous points.
Bill Johnston how do you express the goals more clearly in the Charter. This is necessary because field is new to most Grid people.
Geoffry Fox, need more vision in the Charter. Put into Charter aim of facilitating projects between relevant groups. CG already been some work on certain domains, group could be a repository of ontologies relevant to Grid (either Data or Services)
BJ – where do the working groups show up that will produce standards. At least have a preliminary list of possible WG topics.
BJ not necessary to have another round of comments on charter. Could go to next Steering Group. Can work out details of further discussion via mailing list.
BJ: Need to put a goal into the charter to produce the use cases.
GF: Create an ontology for high energy physics as a Use case.
CG: Has some use cases that can be put off.
Hurricane Use case
Run Mini workshops
.
BJ: Where do the WG’s show up or how can what they produce support this activity. How do we go from here to there, i.e. integration of Semantic Grid into developing GGF standards?
How do we reason one Grid service is like another?
DD: Need to get mailing list sorted out. Action on Dave and Carole.