|
- Updates: Overview and Progress
- Ontology-based browsing and searching
- Coherent file-manager with caching for actors and remote services (see IdentifiersInKepler)
- using LSIDs for identifying actors and services
- relationship to an 'Actor Repository' and EcoGrid
- Pipeline integration tasks
- Actor-actor integration
- Data-actor integration
- How do we annotate datasets, and what can we do with it if we do
- Utilizing Jenny Wang's schema matching tool
- Distributed execution of kepler models
- Choosing one (or more) distributing computing platforms
- Should the Director be kept in the computation?
- Dynamic loading of jar files with actors
- Data access in ecogrid
- query, get, put, etc.
- should these be instantiated as actors?
- archiving kepler products in the ecogrid
- how do we generate appropriate metadata for archived products?
- semantic annotation/registration for EcoGrid data objects
- Actors -- do we have a complete set?
- GIS actors -- more needed?
- Statistical actors
- Anova, regression, various other stats nicely wrapped up (wrap R?)
- Integrating R scripts in a semantically-meaningful way
- Kepler UI
- what's right, what's wrong, what can we realistically fix
- guidelines on how to consistently extend and organize the Kepler UI
- From-scratch design of a semantics-driven UI
- what would we build in a world free of contraints
- Goals and work plan for the next year
- Ontology based browsing and searching
- add functionality to add new actors to the ontology
- add functionality to change ontology views
- add functionality to allow user to put actors into different folders in an ontology
- add functionality to suggest actors that can be used with other actors (SMS)
- add functionality to add ontologies
- change categorization to a connection based cat.
- choose an actor then the org would change to be only actors that would work with the chosen one.
- 3 interfaces
- browse
- search
- contextual reduction of actor base
- Semantic Mediation
- New EML module to describe processing
- data management
- search, organize, store data (morphepler)
- query subsystem
- filesystem management ala morpho
- some see a need to control the file management for controlling ids and file locations
- could specify a directory(ies) that would hold the data/metadata. the user would then drop objects into that area for the system to process.
- system could be searched for kzip (or whatever) files and periodically updated. would need a system wide index of kzip files.
- workflow transfer system
- zip up workflow so it can be easily transfered
- don't require the use of the ecogrid
- archive format
- workflow, data, actors with metadata for each (verifiable-checksum, etc).
- method descriptions for workflows
- extend EML method
- OWL in AdditionalMetadata?
- other possible metadata standards
- ACTION ITEM: need a workflow metadata language for semantically describing
the workflow
- Annotate a specific set of data sources and targets
- start with Matt's Red Spruce integration example (Shawn, Matt, Mark)
- then work through increasingly complex examples (e.g., Cox's biodiv data) (Shawn, Matt, Mark, others)
- Annotate the "component library" so that we know what functions they perform and how they relate to the "conversion" tasks, and how to model the conversion tasks
- e.g., given species abundance and area, "can calculate" density, then label the actors that can do this computation with that annotation
- Finish task list for additional features that need to be completed (Matt, Bertram, Shawn, Mark)
- "XKepler", ability to disassociate UI event stream from execution engine so that the UI (Vergil) can come and go, and the UI subscribes to UI event streams that might be buffered in a real-time system like ORB, and handles messages like GeoVista does for UI events and coordination
- Distributed computation and grid computing
- Two general approaches to be explored
- Kepler interface to existing distributed job system like GT/NIMROD/CONDOR
- New Kepler director that is "distributed aware" and knows how to schedule and invoke remote jobs though one of several services:
- ptexecute on the remote node
- a web/grid service on the remote node
- a managed job on the remote node (cf GT2.x jobs)
- Peer to peer style sharing of computation resources (possibly another grant)
Back to the Annual meeting agenda
|
