-------------------------------------------------------------
DM Session 2 Notes  - Interop Trieste , Thursday May 22rd 9:00 - 10:30
-----------------

By Daniel Durand and Fabien Chereau 

Marie-Lise Dubernet: DM for Atomic and Molecular Standardisation

    History:
    - Besancon Observatory on-line in VO
    - 2004 start collaboration in 2004 with ESAC
    
    Actual Goals:
    Insert Lines Measured, calculated, evaluated
    Insert Observed Spectra
    Implement Some provision for Line Modification (hard)

    Status:
    AMLDM 0.6 - introduction of Physical Quantity
    Data Model for storing lines/transitions from molecular physics/ spectrum

    - Goal is to allow interop: doing cross matching between different  data from different provider
    - Involved with physics community for defining another data model  which is more detailed
    - SLAP is using small number of query parameters on AMLDM
    - Main difficulty is to tag the proper quantum numbers --> has to go  toward AMSdata (www.altova.com)

    Suggestion for evolution:
        - to define the required or not SLAP fields
        - agree on minimal common interface in relation in SLAP
        - try to better match the theoretical/ experimental physicists  community

    Next: DM stable soon
          Xsams on NIST and release 1.0 in Spetember

    2009: DM of line modification
          Query Language
          DM in line in Astrophysical Medium

    Questions:
        - How to query a line at the end?
            concatenation of Qantum number is returned as a string
            link to vocabulary, not user friendly
            client with provide the user friendliness
        - Query language is what?
            Not sure yet
        - Is it planned to evolve SLAP to support the full complexity of the DM?

            There are multiple contexts (theoretical, fusion  physics,  astrophysics) with different requirements. So implement a context sensitive intermediate layer  tuned to the discipline

Francois Bonarel: Observation DM: Status Report

    Need for Observation data model: Provenance, more fine grain characterization classes
    in additionn to Curation, Dataset identification, and Data classes accepted at the  Cambridge meeting (2003)

    Creation of an Observation subgroup

    Concerns: Characterisation data model usage and dissemination is very low  (except for SPECTRUM/SSA)

    Question: How do I publish / use the model
    Question: How do I know what I have to compute to fill the attribute  of the char model

    Use cases for level 4 were identified: 2D response, spectral response,  beam estimate, etc...

    Comment: 2D response is really important (Marc Allen)

    Provenance: Provenance is a concept attached to an Observation that can represent a linkage to its progenitors observations (componants, series, etc...) 
            linkage to original raw data and calibration stuff
            Hard to be panchromatic

    For detailed characterization, trade-off between 
    A) providing only  utypes for describing attributes (easy but not very interoperable) and 
    B) defining them more in details (more complex but more powerful).

    Francois Bonnarel's personnal view: Implementing the second solution, using a minimal description.

    Questions: Doug Tody - Please not make level 2 too complex and add  these new classes as independant entities.
              Could be implemented

           Anita Richard - Level 4 too complex and the most important is a  pointer to the relevant information, aperture mapping?

           ??? - Probably both could be implemented by doing A and moving to B eventually

          Gretchen Greene - Hubble Legacy Archive has good use cases for the  multiple levels of Mosaics

Anita Richards: Data model extensions: Polarimetry and Interferometry

    Small amount of polarimetry data in VO (only NVSS)
    Need to agree on vocabulary and definitions
    
    Polarimetry:
    - Polarization axis: many different ways to describe polarimetry  (Stokes, circular etc..). Associated FITS codes
    - Needs for Vocabulary standards
    - Use cases: Finding (semi-) raw data, Analysing 'Science ready'  polarization images (query on Stokes params)
    - Polarisation axis must be explicitly ordered
    Interferometry:
    - Use cases: Reduced visibility data (pretty easy to manage), raw  data: much more complicated.
    - For many uses, adding a spatial frequency axis in char DM would help.
    - visibility data is required (u,v plane)
    - need polarisation

    Questions: SIA V2 will potentially support polarisation but not  defined  at this moment
    Questions: Tagging the proper polarisation vector is important. Not even converging in FITS yet.
    
    Questions: Andreas Wicenec- VLTI is a small community, just concentrate on scientific extend instead of data modelling
           Igor Chillingarian - Disagree, VO has to enable all datasets to be represented. So then it will be used.

Gerard Lemson: SimDB data model

    Context:
    - online service for discovering and accessing numerical simulations
    - xml representation of DB model storing the simulation parameters
    
    - SimDB instance allows to perform ADQL queries, and VO output. Use  'put','get' for exchanging data
    - Use cases for several institutes simulation resources.
    - Use relational DB to store data and allow for ADQL. Use custumized xml  I/O to handle complex data
    - DM itself expressed in UML (xmi representation) -> Generate  everything from it automatically
    - The framework allows to describe very different data, because each simulation produces its own specific data but defines concepts and  vocabulary.
    - Future: move to recommendation track.
    Q&A: Code hosted on googlecode ('volute' project)