Ticket #764 (new task)

Opened 15 years ago

Last modified 14 years ago

Incorrect citation in BODC DIFs

Reported by: rkl Owned by: kthorne
Priority: required Milestone: NDG2 Cleanup
Component: community Version:
Keywords: Cc:


The BODC DIFs have BODC as the originator in the dataset citation element, not the data originator, which is what it should be. Fixing some of these will be easy, but what we do about some of the others (e.g. the NODB CTD dataset) will need a bit of thought and/or discussion.


Source_Name.htm Download (5.8 KB) - added by kthorne 14 years ago.
GCMD DIF writers guide for Source_Name

Change History

comment:1 Changed 14 years ago by kthorne

  • Status changed from new to assigned

comment:2 Changed 14 years ago by selatham

  • Component changed from T01_Discovery to T10_Working_Grid
  • Type changed from task to metadata
  • Milestone changed from PROD to PROD Final

Changed 14 years ago by kthorne

GCMD DIF writers guide for Source_Name

comment:3 Changed 14 years ago by kthorne

  • Status changed from assigned to new
  • Owner changed from kthorne to lawrence
  • Type changed from metadata to task

Hi Bryan,

To fix the ‘Sources’ area always displaying BODC, in the Discovery record, I updated the Data_Set_Citation> Dataset_Creator field for our DIF’s and resubmitted them in July 2007. See extract from an edited DIF below:


<Entry_Title>AUTOSUB Under Ice (AUI)</Entry_Title> <Data_Set_Citation>

<Dataset_Creator>National Oceanography Centre, Southampton</Dataset_Creator> <Dataset_Title>AUTOSUB Under Ice (AUI)</Dataset_Title> <Dataset_Release_Date/> <Data_Presentation_Form/>

<Online_Resource> http://www.bodc.ac.uk/data/information_and_inventories/edmed/report/1048102/</Online_Resource> </Data_Set_Citation>

However, I notice that this has not remedied the Sources element in the html view of the Discovery Record, as this record still lists BODC.

I notice from BADC’s DIF’s that you have used the Source_Name tag. However, when looking at the GCMD Dif writer guide (see attached file) this in not appropriate for the BODC DIF’s as we are referring to organisations rather than platforms. Hence our DIF’s do not contain any Source_Name tags.

Do you plan to edit your code to display the Source_Name elements and/or the Dataset_Creator elements in the ‘Sources’ area?

comment:4 Changed 14 years ago by lawrence

  • Owner changed from lawrence to rkl

I don't think this is a problem for us to solve.

What we (NDG) are doing is using moles to map the observation station into source, the data production tool into sensor, and the activity into the project. I think that is the Right Thing (TM) as far as NDG goes. I don't really care what GCMD Think.

What is happening for MDIP ingest I'm not sure.

What BODC does is up to you. I think.

Can we close this ticket?

comment:5 Changed 14 years ago by rkl

  • Owner changed from rkl to kthorne

Hi Bryan,

I fully agree with one of these, namely DPT to sensor. I also think we can get away with activity to project at the dataset (but not data granule) level. It's the mapping between observation station and source that doesn't work for most of our data. For 99% of our stuff this means that the 'source' will either be a handful of research vessels or a scattering of localities on the surface of the earth. How one can get the identity of the dataset source from any roles associated to these totally escapes me.

Right for 'NDG' to me reads as 'right for most of the data held by BADC', which p* me a bit. The other thing that I will moan loudly about is that this is the first time I have been told about this mapping between MOLES objects and DIF elements. It might have been information known around RAL, but it didn't diffuse outwards. We've been trying to design holding hand-coded DIFs without this insight, which has been a significant handicap.

As a quick fix putting the necessary information into 'source' in our holding DIFs is a possible way out for the short term. So, I'll reassign the ticket to Kay to get this done.

However, Bryan, I think the main point to come out of this is that the guts of our discovery system is based on mappings (DIF to MOLES, MOLES to DIF) that we have to determine by detective work matching inputs to outputs. If I could have these in documented form then I could look at the way we're populating MOLES and make sure in the longer term that dataset DIFs we produce from MOLES deliver correctly.

Note: See TracTickets for help on using tickets.