|
General agreement that collaboration would be productive on a number of issues. Some potential areas for collaboration:
- Develop common grid services architecture for describing components and workflows
- Execution models
- Cycles
- temporal domains for execution (Director in Ptolemy)
- Can SEEK adopt a GUI for workflow composition?
- can we use componenets from multiple environments
- can the GUI realistically be deployed to our target science community?
- IP issues
- financial issues
- computational steering for distributing and controlling execution
- How can KR and semantics be used in an open system for driving the analysis process?
- using semantic mediation in various system components (data integration, workflow composition)
Some choices: XPDL, DPML, MoML, (Monarch), PMML, VDL/Chimera
- need to ask Victor and Liying about benefits and problems with particular languages from a verificatin/validation/static analysis perspective
- Task assignments
- Workflow languages
- Review and summarize features of relevant workflow languages (Bertram)
- determine if DiscoveryNet and GV Studio have explicit execution models, and if they are exposed through their respective workflow languages (e.g., DPML)
- create static representation of steps that can be bound at runtime to implementation classes, and that can be created by the scientist before an implementation has been wrapped or discovered
- adapt Ptolemy for use by BEAM to more rapidly create some more workflows for real analysis scenarios that we can analyze for patterns
- how can we add more informationon to the link to facilitate tranformation
- Explore GeoVista Studio
- use it to create an interesting visualization for the GARP pipeline (ask for assistance if needed)
- figure out execution model
- determine if an executing Ptolemy workflow can output to GV Studio for visualization
- follow up on extraction of ontologies from scientists and where GVStudio is at on linking concepts to data
- Explore data typing
- how can we make this WSDL compatible, and specifically xsd data types?
- determine how do we link semantic types in MOML and other languages? probably a new property
- Follow up with DiscoveryNet
- what are our requirements for collaboration? is source code required? is open source required?
- establish conf call with Yike Guo when we have our plans worked out
- Explore flexible generation of transformation steps
- Examine Blue Titan approach to transforming WSDL steps
- Figure out how SMS/AMS can help discover steps that fill a particular transformation role
- Figure out how to generate GUI for creating first-class transform steps for particular mapping purposes
- includes schema tranformations
- this can use the target semantic type as the target view for transformation
- determine pros/cons of having transformations directly associated with links versus being implemented as a step unto themselves
- Contact Ptolemy people
- Analyze case studies for patterns and "correctness"
- Quick and dirty pipeline editor that allows construction of test cases even if not executable (modified Ptolemy)
- Analyze case studies for patterns and "correctness"
- Identify key collaborators and bring them to All-hands meeting
- Figure out long-term conceptual approach to data integration (e.g., as described by Villa)
|
