Kr Sms Taxon 2005 Mar 8

This is version 5. It is not the current version, and thus it cannot be edited.
[Back to current version]   [Restore this version]

Breakout Session for KR/SMS/Taxon 2005 March 8

Participants: Aimee, Bertram, Shawn, Dan, Mark

Possible collaboration point: reasoning over functional attribute database

Semantic mediator will call TOS

EML

• Maintain GUIDs in metadata or resolve dynamically
• Could bind resolution if done by original data collector/author
• Third party resolution results could improve over time until TOS cache is somewhat stable.
• Who is responsible for the association if not maintained in EML?

SMS focusing on following scenario: SMS WILL construct for a given (number of) dataset(s) the best set of information (limited/query concepts; as “complete a filled-in concept as possible”)) to feed to the TCS (binomial, year, location, e.g.)—and a specification of the grain to which you want these resolved (default to finest grain?). TCS then returns (a potentially integrated or presentation of possible integration options) of these two lists of concepts. Aimee says that Taxon is aware of this, but hasn't gotten to it yet, as it is the hardest/most complex function (and builds on the others).

Use Case:

• Scientist brings in 2 datasets
• Resolve names for each list
• Can this data be integrated? yes/no
• Need Integrate function (This is already on the task list for Taxon)
• integrate(list1, list2, criteria)
• Need Compatible method:
• compatible(concept1, concept2, criteria)
• compatible(list1, list2, criteria)

For a more automated workflow approach - maybe associate taxonType as output port for dataset actor and input port for analysis actor

Additions/Modifications to TOS:

• Additional input for findConcepts:
• ancillary data to help resolve concept (i.e. geographic, functional, temporal data about dataset)
• level of desired resolution (could be part of algorithm)
• Automatically resolve a list of names with different resolutions/ranks to a rescaled list of concepts (lump to reconcile with other list, criteria)
• step 1: names to concepts (with user interaction?)
• how is this determined when taxonomists make judgements based on taxonomic author/provider?
• step 2: concept to parent
• third party relationships or originally defined concept relationships
• Compatible(concept1, concept2, criteria) - someone needs to define criteria for analysis.
• List of names
• are these most recent/best names
• what are ambiguities
• need for human intervention to resolve
• which GUID associated for co-referencing with other data.
• Relation with EML Taxonomic coverage: if binomial, and launch TCS search, how capture results, especially if these required human intervention; is TCS resolution going to be associated with EML as semantic annotation.

SMS tasks:

• SMS define criteria/ontology for ancillary taxonomic information
• Taxon redefine API arguments based on above