2012Feb08

From FilteredPush
Jump to: navigation, search


Etherpad for meeting notes: http://firuta.huh.harvard.edu:9000/FP-2012Feb08

Agenda

  • NSF Workshop RFP
  • AOD
    • Progress on paper
    • Annotation constraints as owl/Annotation generation
    • Response Annotation
  • AnnotationProcessor
    • Integration Progress
    • UI Design
  • API For Query and Cluster Finding
  • Roadmap next steps:
    • Driver able to handle new determinations and update determinations.
    • Refine WebUI for annotation processor with user input.
    • Generate response annotation from annotation processor
    • Create annotation from Morphbank
  • Pending Tech Issues
    • FilteredPush Requirements Report_on_FP_Requirements
    • Tech group needs to make a decision on or set a date for decision for the query language for pub-sub for Apple Pie.
    • Tech group needs to make a decision on or set a date for decision for the domain objects supported for Apple Pie.
    • Tech group needs to decide on or set a date for decision for the scope, composition, and implementation of the "global cache."

Reports

Notes

Filtered Push Team Meeting 2011 Feb 08

Present: Bertram , Lei, Jim, James, Bob, David, Paul, Maureen


  • NSF Workshop RFP

Bob: Nothing back yet.

  • AOD
    • Progress on paper

Bob: Comments back from Paolo, integrating. Very close to being able to submit.

Paolo has put up a new version of AO, preliminary. Flattened everything except provenance.

Bertram: Is this related to the question about how we look at this stuff?

Bob: For the AO ontology itself there are visualizers, issue is when we produce RDF instance documents based on an ontology and want to visualize the individuals. Can point at suitable visualizers for OWL classes.

    • Annotation constraints as owl/Annotation generation

Bob: Put on wiki http://etaxonomy.org/mw/ApplePie_Supported_Annotations preliminary ontology that expresses constraints on standard sorts of annotations we would make (David currently working on generating new determinations/correcting existing determinations). The typical concerns of ApplePie will reuse these. Lei has rules for expressing these in Swirl, parts of these are easily expressed with restrictions on classes in OWL DL. Annotation subtyped by two ApplePie annotation Types.

Lei: How do you want to use the SWRL rules?

Bob: I have several things in mind: (1) validation, enforcing rules in reasoning not query (2), should be able to extend this to configuring a network instance by generating code for annotation processing using Jena class generation.

Lei: Switching now to using sqarqul for validation, Swrl works best with reasoning, harder to use for document validation.

Bob: Objects constructed locally in RDF, then associated with something global, dcterms:references is in play (Paul: coming from iDigBio driven examination of GUID references in DWC).

Bob: Examine RDF: Where in the RDF does it tell me what record in the database I am expected to change, but there are no formal semantics in the RDF or owl to base the decision of what to do, thus in code there will be a decision to use the object of the dcterms:references, but that isn't something the code can deduce. Code will start acreeting things coming out of your head.

Lei: Annotation has darwin core tripple, problem is asking which record this refers to?

Bob: yes,

Maureen: You want your annotation documents to do two incompatible things: (1) describe anything that anyone could describe, and (2) have very specific uses of annotations within ApplePie relating annotation content to specific database actions.

Bob: Exactly why I am proposing subclasses of annotations.

Lei: Currently using sparql query to validate and get information from (type) the annotations.

Bob: All I'm saying is that if the sparqul query is based only on the type/class of the annotation, then the code should be more resiliant to change - the typing goes into the ontological constraint set, not the query.

Maureen: My experience in writing out these stubs is that all you need to get local types and an identifier.

Bob: We are on the same page

Maureen: Amplifying: The driver needs the table name and primary key value (on the local side), on the other side, provide an object and identifier that the driver can resolve to the table and PK.

    • Response Annotation

http://filteredpush.svn.sourceforge.net/viewvc/filteredpush/FP-Design/trunk/ontologies/ao/aod_example_response_annotation.rdf?revision=940

Paul: Brief description.

James: Like general direction.

Bob: Characterising as annotation dialog.

James: May be a DarwinCore term to carry the sense expressed within the asText of the Evidence.

  • AnnotationProcessor
    • Integration Progress

Maureen: have new version deployed on firuta. Have the UIs integrated, working on integrating the links between the UIs and the underlying functionality.

Lei: Back end logic for annotation processing is done. Good path for combining duplicated functionality.

Maureen: Code for generating annotations in svn?

David: In svn is one step back, about to check in latest version.

Lei: Multiplicity? One topic per annotation, or allow more than one.

Bob: ApplePie, allow multiple topics until we know where the problem is.

Maureen: Value?

Bob: If someone has more than one thing to say about an object: (1) you can get race conditions, (2) only other route to avoid is to create two annotations in one annotation set. What is the problem?

Maureen: Makes the UI more complicated? Also if there is an atomic thing that has to happen, why can't it be represented as one thing.

Paul: Scenario for a race condition.

Maureen: Deeply skeptical.

Bob: Interface separates what you are commenting about from what your comments are: Could provide list of suggested changes to.

Lei: What are the relationships between multiple topics for one subject? Order may be important.

Bob: None of this has to do with how you model assertions about a single thing. If order matters then the annotation processor has a problem.

Maureen: Perhaps the solution for the sequencing issue is a sequence.

Bob: This may be hard in RDF. Will investigate. (See Sequences_in_RDF )

    • UI Design

Maureen: Something for James/Jim to review in next week

James: Mouse overs in UI? Part of in this?

Maureen: retaine.

  • API For Query and Cluster Finding
  • Roadmap next steps:
    • Driver able to handle new determinations and update determinations.
    • Refine WebUI for annotation processor with user input.
    • Generate response annotation from annotation processor
    • Create annotation from Morphbank

Reviewed, ok.

  • Pending Tech Issues
    • FilteredPush Requirements Report_on_FP_Requirements
    • Tech group needs to make a decision on or set a date for decision for the query language for pub-sub for Apple Pie.
    • Tech group needs to make a decision on or set a date for decision for the domain objects supported for Apple Pie.
    • Tech group needs to decide on or set a date for decision for the scope, composition, and implementation of the "global cache."

Notes:

Bertram: On another note: I chatted with Lei and James re. a little Kuration workflow paper for this event: https://kepler-project.org/developers/kepler-workshops/iccs-2012-advances-in-the-kepler-scientific-workflow-system-and-its-applications The deadline is soon though (Sunday). If you're all OK with this, we'll go ahead (with James taking the lead on the Kuration story broader vision ;-) and try to put this in by Sunday..

James: Kepler workshop soliciting papers. 2-3 page short paper as a presence with vision and solicit some feedback.

Bob reminds us to use the prefixes and entity declarations on: http://etaxonomy.org/mw/Namespace_Convention_in_AO