This is an attempt to pin down the what will be required for the demo mainly from my point of view after discussion with Chris, Kevin and Mina. It is presented as the resources needed for the demo and a rolling list of TODO items.

-- ChrisWroe - 17 May 2003

1) Workflows that will subsequently found and run. Peter, Tom, Darren etc seem to have that under control.

2) Annotation of extra metadata to workflows to

1. 1. allow the workbench and gateway to hook up its data to the start of the workflow and appropriately annotate the results of the workflow. (historically called a job.xml file in the gateway)
   2. allow users to search for the workflow.

Nottingham and Manchester have agreed to put these metadata in a single place as they significantly overlap.

• TODO A - develop a schema for this file. - done in CVS. can now be created using PEDRO
• TODO AA - document schema with PEDRO help files.
• TODO B - develop code to prime this metadata file with information in preexisting scufl, wsdl and service description files.

3) Publish a description of the workflow into the registry view. This will enable the find service to look for it in the same way it would look for services.

• TODO C - resolve exactly how the workbench will get hold of a view and keep hold of it.
• TODO D - develop code to take the workflow metadata file and perform a service registration over the view API.
• TODO E - finalise the schema of the DAML+OIL class expression to describe the workflow. (It must be the same as that for services)

4) Add data to the mIR for which we need the data itself, and a metadata annotation to allow: a) the workbench to know how to display the data b) the find service to narrow down the search of services and workflows to only those that accept the same semantic type as that of the data chosen.

• TODO F - link classification data wizard to the ontology of semantic data types
• TODO G - finalise where this ontology can be found, and what format(s) it is available in.
• TODO H - may need an ability to do this annotation or change it after the data has been added to the mIR.

5) pick data, view data

• TODO - ?? Nottingham todo list

6) pick workflow using a wizard. The workbench will pass the find service the concept type of the data chosen. The find service will need a reference to the view.

• TODO - view factory within workbench? see TODO -C
• TODO I - concept type passing within the wizard - may be done

The find service will return a service key which will need to be related back to the mIR key for the workflow definition.

• TODO J - code to query the view to return the mIR key

7) The semantic description of a workflow will be partial and so it will be pretty essential for the user to be able to look at the workflow. One simple option would be to allow the user to look at the dotty generated png file from taverna.

• TODO K - png viewer?

8) Once chosen the workflow is run. On completion the data is written into the mIR with appropriate metadata annotated to the data (such as concept type)

• TODO L - code to extract relevant metadata from workflow metadata file.

9) View results and provenance?

• TODO ?? Nottingham todo list

The search for workflows wont look very convincing if there are only a handfull of them. Options are to include services in the results even though there will be no facility to run them. or to automatically write a set of single step workflows to wrap the SOAPLAB services and register them. Either way it will be nessary to associate semantic metadata descriptions to the SOAPLAB services that Martin will register.

• TODO M - work on a strategy to key in our semantic descriptions of EMBOSS services with Martin's service registrations.

• TODO N - extend ontology to encompass worflows and the services used in those workflows.