search

binary-array-ld / netcdf-ld

netCDF-LD
1 stars 2 forks source link

Encoding Linked Data information directly into the netCDF metadata #3

Open lesserwhirls opened 7 years ago

lesserwhirls commented 7 years ago

Greetings! I was wondering if part of this project will involve the creation of a standard for directly encoding Linked Data syntax into the netCDF metadata. The RDF/Turtle example output looks pretty close, but not quite a direct representation of the how things might look in, say, a cdl. Thanks you!

jyucsiro commented 7 years ago

Hi, on embedding RDF in netCDF, we discussed this in a previous meeting but didn't pursue the idea as the group thought adding info and context could be achieved outside of the netCDF data itself via Linked Data. However, keen to learn about your thoughts on it and a use case for embedding it in the netCDF itself.

lesserwhirls commented 7 years ago

Let's try to setup a telecon to discuss netCDF-LD with an eye towards embedding RDS in netCDF, the idea of which would to be able to cleanly support multiple metadata conventions in netCDF (any version).

Doodle poll for those interested (first shot - let's see how it goes!). Make sure the poll has your correct time zone set. Doodle tries to use IP address to auto select it, but it does not always work 😢

An idea of the time zones involved here 😀

Please tag anyone that you think might be interested.

lesserwhirls commented 7 years ago

@ethanrd @DennisHeimbigner @cwardgar @WardF

lesserwhirls commented 7 years ago

@dopplershift @jrleeman

marqh commented 7 years ago

hello @lesserwhirls

there are a number of participants on the timing poll now. I see from the times that there is a useful window on Wednesday 10th 2100 or 2200 UTC

As that is tomorrow to me, may I suggest we have a swift Go|NoGo decision on this ticket as to whether to forge ahead with this time?

The next available dates are in the week 22nd May. We can target one of those instead, if that is preferred.

@lesserwhirls @jyucsiro are one of you aiming to host this meeting?

i'll be prepared to dial in tomorrow evening, in case that's the preferred option. If tomorrow is off, please may we target a specific date in the week commencing 22nd and get it in my diary?

thank you

lesserwhirls commented 7 years ago

Hi @marqh, all:

Thank you all for participating and being flexible 👍 Let's target Tuesday 23 May 21:00 - 22:00 UTC

https://doodle.com/poll/ai2att3zyxrs6e5m

I'll get us setup with a meeting place online (likely webex) and will post the details once I have them. Thanks again!

jyucsiro commented 7 years ago

@lesserwhirls Thanks for figuring out a time to meet! Yes, send through an invite. cheers.

lesserwhirls commented 7 years ago

Invitations to the WebEx meeting for tomorrow were just sent out. Let me know if you do not receive one. Thanks!

jyucsiro commented 7 years ago

Thanks. Got it.

marqh commented 7 years ago

@lesserwhirls

I am afraid that I have been unable to log into the WebEx meeting

I am getting strange errors during the connection process and I cannot find a way around them to join you

I am sorry

huard commented 5 years ago

New to this, but interested in knowing the outcome of this discussion.

My use case would be to embed provenance information in a netCDF file. At the moment, provenance is captured in text form in the history attribute. In practice, my experience is that this solution does not scale very well and gets confusing when a netCDF file is derived from multiple different files through complex workflows. The history attribute just becomes a mess and is not standard enough to be understood by a parser.

I was thus interested in generating an RDF document embedding the full provenance of the data and storing this as a netCDF global attribute (see metaclip.org for an example of such provenance information embedded in a graphic).

There are examples of such approaches in the wild already (ESMValTool, page 200), but the CF-Convention does not mention "provenance" at all.

Suggestions on how to move this topic forward ?