ObsTAP.creator_name multiplicity

[Guillaume Mella]

Le 09/12/2014 17:09, Douglas Tody a écrit :
> By aggregator I meant merging Obscore records from multiple sources into
> a single database, which sounds like what you are doing.
More or less. That is the concept, but we have to convert various source 
information format into obscore records.
>
> Your question refers to multiplicity, however I think there may be an
> issue with the use of creator_name.  This is supposed to be the name of
> the entity that created the dataset and defined the CreatorDID if any,
> e.g., the observatory, survey project, advanced data service, etc. that
> created the data product (this predates Provenance with later duplicated
> and expanded upon some of this information).  Not the PI for the project
> the observations / data products represent; that would normally be part
> of the provenance, linked to proposal_id, probably via a different
> table, which would allow for describing the PIs in appropriate detail
> (this is not currently addressed by Obscore).  Part of the motivation
> for creator_name was to provide information about the origin of the
> dataset, separate for example from publisher_id which is usually
> something completely different.
>
> Hence, there should not be a multiplicity issue for creator_name if it
> is used correctly.
>
Ok, we probably are misusing this field.
> If the creator_name values input are incorrect a possible solution would
> be to correct them in your aggregated / merged table, and identify the
> original source of the data.

Thanks a lot for your help, ideas and clarifications.
--
Guillaume

>
>      - Doug
>
>
> On Tue, 9 Dec 2014, Guillaume Mella wrote:
>
>> Hi,
>>
>> Sorry for the confusion, there is no aggregating mecanism on data. At
>> the present time, we do not process data and only have to register in
>> the same database single observation references coming from multiple
>> sources.
>>
>> We encountered this case when we asked for the PI information per
>> record and get two people as result.
>>
>> Final question could be:  how to fill the creator_name for a
>> dataproduct associated to a program handled by two persons ?
>>
>> Cheers,
>> --
>> Guillaume
>>
>>
>> Le 09/12/2014 16:03, Douglas Tody a écrit :
>>> Hi Guillaume -
>>>
>>> As Mireille mentioned it would help to know more about your use case.
>>> It is not clear if this is merely an aggregator, aggregating multiple
>>> observations from external sources, or if you are actually producing new
>>> data products from multiple external sources.  It is the latter case I
>>> addressed in my previous email.  If one is merely aggregating records,
>>> these would be separate records, with different obs_ids, in an ObsTAP
>>> index.
>>>
>>>      - Doug
>>>
>>>
>>> On Tue, 9 Dec 2014, Guillaume Mella wrote:
>>>
>>>> Hi,
>>>>
>>>>  We are building a portal of observations from multiples facilities on
>>>> top of ObsTAP.
>>>>
>>>>
>>>> One of our issue is that given some inputs we would have to ingest
>>>> some records with multiple names in the 'obs_creator_name' column.
>>>> But the multiplicity of this field is not addressed in the ObsTAPV1.0
>>>> doc (B3.1 and B3.6 are duplicated).
>>>> This column renamed 'creator' in VODML also get a 0..1 multiplicity
>>>> looking at
>>>> https://volute.googlecode.com/svn/trunk/projects/dm/vo-dml/models/obscore/ObsCore.html#DataID.Creator
>>>>
>>>> .
>>>>
>>>>
>>>> As a first workaround, we plan to use the same convention proposed in
>>>> the creator_seq column of the RegTap.
>>>> If it is a wrong choice, why and what would be the proper guideline to
>>>> follow ?
>>>>
>>>> Cheers,
>>>> --
>>>> Guillaume
>>>>
>>
>>