Etherpad for meeting notes: http://firuta.huh.harvard.edu:9000/FP-2011Sep13
- TDWG Demo
- Annotations and the Mapper
- Architecture Implementation
- GRDI timeline for FP network installation
Bob- Cleaning of AOD examples. Recode Example 4 to DwC, with two slightly different proposals.
- Submitted Apple Core abstract for TDWG lightening talk which was accepted
- Discussed with Tim about the mapper architecture
- Studied and discussed AO to figure out what kind of database update request will be expressed in an annotation
Maureen: Question: How to support subscriptions on any topic, without for each subscription talking to each data provider.
Paul: JMS model for subscriptions: (1) annotation plus payload launched into network. (2) annotation results in query to network data providers (3) result of query are bound to message as key value pairs in the message metadata along with key value pairs for the annotation payload values (4) expressions of interest are framed in terms of key value pairs by subscribers. (5) message handling system takes a message and compares key value pairs in a message to interests, and forwards message to anyone with a match.
(Further discussion, didn't catch.)
Annotations and the Mapper
a) Can we use dwc:Taxon instead of applecore:taxon just as the to:TaxonName is used the example 3? In this way, the dwc:Taxon is a data property instead of a class and its subject is a freeText.
(it's now in a different namespace for reasons we can talk about tomorrow. it is now fp:taxon. That doesn't impact your questions though)
b) Why do we need both applcore: taxon and the dwc:ScientificName? I guess the reason is the same as why both to:identifiedToString and to:taxonName are needed. So the applecore:taxon is a structured information while the dwc:ScientificName is a free text string.
The problem in general is what to do when there is a "string" way to say something in an annotation and a "structured" way.
Should we say in this instance that when there's a string and a structure representing the same thing, the string trumps the structure?
How are clients supposed to construct annotations that represent botanical duplicates? One solution is to create owl documents with the relations of the various classes. People will need help doing that.
The problem is that there is no standard DWC Owl representation. Are options for the mapper are to use DWC understood as XML schema ("the mapper understands DWC"), or do we mean an Owl vocabulary, or our own effort to build an Owl representation of DWC.
Bob's position is that in a few weeks of added examples we'll find out how many relations we would need to add in that last case.
We know that we will need sophisticated relationships for the queue filtering, this is the one place that the flat xml DWC would be insufficient, but not for the mappper.
on the agenda for Friday Along with mapper/annotations.
GRDI timeline for FP network installation
James: Large proposal in Canada, 4+year project with timelines in proposal. FP current funding over before GRDI deploys it (2013). Hoping to have software and documentation to deploy instance in Canada.
Jim: What is meaning of safe date?
James: Date by which there should be product from FP grant allowing creation of a deployable instance.
Paul: Only concern is requirements for security,
Bob: Concept is partly to delegate some of security (authentication/authorization) to DataOne.
James: Happy to include mention of coordination in 2012 for security requirements.