Hack-a-thon Sessions

• Markus D. : how can we realize the simple thing : time series —> progenitor

need some description of an Activity : Extraction of a mag from an image etc .. follow the Prov relations : wasDerivedFrom or wasGeneratedby + Activity ID+ Used

• Carlos's use-case : Provenance from points in a SED or light curve

linking points in a data set to other dataproducts : extracted image, extracted spectrum, etc. Is it simple to solve? -without the prov DM: datalink + progenitor link if it exists datalink + preview if we want to illustrate only the point via an image -with provenance info : select the wasDerived relation in Prov : it will provide you with a progenitor if present.

• François : linking the prov info from Obscore : a prov column which contains a link to a Prov document in one PROV format (PROV-N, or PROV-JSON)

• IVOA PROV Data model design :

A suggestion to handle ActivityDescription by Ole

PROV-N document https://provenance.ecs.soton.ac.uk/store/documents/118181/

Example of Ole:

• Top graph: representation of an activity executed on real data
• Below graph : Define a template for a an ActivityDescription here “musesci_post” with known roles, with context dependent names, for instance here in the muse:namespace

a Bundle class (W3C compatible) for expressing ActivityDescription - solves the pb of roles A realisation of an Activity:

• uses the activity template
• uses entities with used roles and generated roles defined in the bundle for description

Musesci_postT1A : Activity Instance has relation “Used” with role= “voprov:description” with instance Musesci_post:Bundle

All other “used” relation should instantiate a role defined in the related Bundle with domain specific namespace, here “muse:” Bundle is a W3C class derived from an Entity. This should be also possible in a VOTable serialisation

Comment François :

check that this can be coded into the VOTABLE serialsition as well.

Suggestion Michèle :

apply this template solution to a Pollux activity example, using 2 chained activity templates.

Comment Mireille:

+ by this construction we can gather in just one construct the description part of an Activity , together with the roles ( formally defined as attributes of UsedDescription) + the roles at execution can be validated agains the activity template .

Discussion with Marco Molinaro about the potential use of the TAP-Autocomplete library at INAF (Trieste). Topics included a description of the available modes of usage, and what would be required at Trieste for potential installation and use of it. Carlos Rodrigo Blanco also expressed interest, and proposed the idea of having the library as a browser add on, which would work on any TAP GUI.

Inclusion at INAF would probably work better if the TAP-Autocomplete library is used inside a page that can be served in the same location of the TAP base_url. Since INAF in Trieste is starting using Gregory Mantelet's TAPlib that already sets up that pseudo-frontpage for its TAP instances and also considering that TAP-Autocomplete uses Gregory's adql.js library, then it looks more useful to have the efforts merged.

Collaboration on this will go on among Stelios, Marco (and Sonia), contacting Gregory and possibly using the votable.js from Andre Schaaff.

After TAP-autocomplete discussion, with Dave Morris joining in, some discussion took place about the Firethorn application and its potential usage to deploy a global TAP service on top of existing datacollection-limited TAP instances at a data center.

The Hack-a-Thon discussion about this topic were presented in a talk during the morning session of Day 2: multi-d-dal-feedback.pdf (François Bonnarel/ Marco Molinaro)

• DataLink discovery outside ObsTAP/SIAV2 context : Resource descriptor, ucd, LINK element and Access GROUP discussed
• DataLink semantics list extension
• DataLink for local resources
• Resource Descriptors processing difficulties
• SIAV2/SIAV1 services combination
• SIAV2 for virtual data missing
• SIAV2 request for list of available collections, instruments, facilities
• SODA lack of implementations. importance of semantics and description for SODA lines in DataLink.

The effort is led by TDIG WG with inputs from DAL, DM. Application is getting in.

Strong involvment of ASTERICS partners.

Discovery metadata, TimeFrame details and Time representation reached a consensus

Data Modeling of TimeSeries is designed as extension of CubeDM. Work by MCD and ML. convergence reachable.

Serialization and representations: several attempts with several methods (VO-DML and utypes) on the same sample dataset have been tested. VizieR has developed a prototype implementing these various proposals Hopefully progresses will be made in VIctoria interop for convergence.

Discussion in the hack a thon was also focused on how applications (TopCAT, SPLAT, TimeSeries) can ingest that. Edimburg group has the intention of developping a TimeSeries specific tool to ingest that and process these data.

Mark T and Markus D, Mark T and Carlos - useful requirements input from data providers

Marco M and Markus D discussed about the reasons to put a modular solution in place, connected risks and potential benefits. Manageability increased costs versus portability and distributed developing effort were considered.