W3C

Spatial Data on the Web WG Coverages Sub Group

18 May 2016

Agenda

See also: IRC log

Attendees

Present
ByronCinNZ, billroberts, dmitrybrizhinev, duo, sam, phila, Maik, Rob, Atkinson, jonblower, kerry
Regrets
ScottSimmons
Chair
billroberts
Scribe
kerry

Contents


<scribe> scribe: kerry

<billroberts> https://www.w3.org/2016/05/04-sdwcov-minutes

<scribe> scribeNick: kerry

<scribe> chair: billroberts

proposed: approve minutes https://www.w3.org/2016/05/04-sdwcov-minutes

<phila> +1

<billroberts> +1

<Duo> +1

<Maik> +1

+1

RESOLUTION: approve minutes

<billroberts> https://www.w3.org/2015/spatial/wiki/Patent_Call

<billroberts> https://www.w3.org/2015/spatial/wiki/Meetings:Coverage-Telecon20160518

Brief recap of previous meeting

<roba> thanks bill - will provide a very quick statement of intent

billroberts: handy template for FPWD

Updates on example implementations: ANU

<Duo> http://anu-linked-earth-data.github.io/#/

Duo: basic demo at link
... very simple client app, have taken a small landsat subset from agdc and loaded to jena and built client app to access data as rdf

<Duo> https://github.com/ANU-Linked-Earth-Data/ontology

Duo: data currently loaded dynamically as RDF datacube in Jena
... 2nd lin is ontology repository, shows datatime, resolution, band etc

,,,client app does basic time-based search over tiny data, can click on a particular area

scribe: also a link button to link to data itself, but now a dummy
... looking for some more data and more complex queries and multiple bands

jonblower: very nice, but didn't get is ontology describing pixes or a big binary blob?

Duo: binary blobs becuase pixiels are too big -- we tested this and discussed at an elalier meeting

<Maik> Sample query

billroberts: can we ss the rdf?

duo: see query and json result here
... maik has posted [what?]

billroberts: please post a link to the rdf

roba: looking at ontology on link .. has a lot of axes concepts.. how does this realte to rdf datacube -- it is different

dmitrybrizhinev: it inherits from rdf-datacube -- are you looking at right link?

roba: maybe not

<phila> I think it's https://github.com/ANU-Linked-Earth-Data/ontology/blob/master/ANU-LED.owl

<Maik> (my link works, just copy paste it in full, don't click on it)

dmitrybrizhinev: you need to look at anuld

billroberts: will take a proper look afterwards

Duo: on top of that we had a chat with rob woodcock (AGDC) about uses of link data and coverage, and a similar chat with ed parsons
... Rob wants us to be smarter about queries e.g summer above tropic of capricorn
... or every second month
... does not like resampling becuase sicentists don't like it

<roba> I have already captured these alternative use cases here: https://www.w3.org/2015/spatial/wiki/Data_cube_for_coverage

Duo: also meeting at ANU Fenner school wher eenvironmental data on a map was broken into "meaningful" regions
... ed wants to see restful APIs over SPARQL, things that startups can access

phila: wnat to use some of this for the Chinese project report
... if I were not an rdfhead what is the point?

duo: still early stages... describing data in a meaningful way.. by having it here can be easier to access.. heading towards resful approach without sparql

phila: i see a column with diagonal black stripes -- what is this to a non-expert?

Duo: this is an artifact of lansat -- many people want this - it is in the data
... if you don't want this this may not be what you want

phila: want to tell people this is a good way to stitch them together

jonblower: what do users need is a good question? yes scientists often want it close to original, but others want the over visual
... good to bring these issues out

bill: creates rquirement for metadata

roba: i know believe that we may not want to turn all inot rdfdatacube but this is a useful test of the competency of the model

<phila> Kerry: There's a deeper question in there. The RDF QB is only appropriate for moderately well developed imagery as you're assuming it's laid out on a regular grid.

<roba> rdf-QB does not assume a grid. its agnostic about dimensions

<phila> ... There's a slightly deeper problem, it's not juyst about what has been done to the data. It's also about how far we go in saying what the satellite image shows.

<phila> ... I'm happy with this though

<phila> ... What is an appropriate data model for represetning coverage in general.

<phila> Kerry: Not all coverages map into easy data models

jonblower: convenient that AGDC is thinking about what level of pre-processing to do

billroberts: we are going to have to deide if we need a suite of alternative solutions for say unprocessed towads that has been through a degree of processing

<Zakim> phila, you wanted to make a quick aside about https://www.w3.org/TR/2016/WD-dwbp-20160519/#EnrichData

phila: discussion abouit various access to dat reminds me that DWBP doc that will be last call tomorrow and asking for reviews
... there is a BP on "data enrichment" and you need to say which value are processed and which are raw.
... hoping someone will look at this

roba: my undertanding that rdf datacuibe does not require regular dimensions..i am looking at gridded things

billroberts: yes, may be one-dimansional eg all the towns in australia

jonblower: th3e other thing about users is that they can go from one type to the other but they need an entry point to start low and help people to ramp up
... by linking between representations and processing steps

University of Reading/CoverageJSON

<Maik> https://lists.w3.org/Archives/Public/public-ceo-ld/2016May/0000.html

Maik: as on the CEO-Ld mailing list have added how to derive new data from existing data
... also how you can tile things over any dimension in our format

<Maik> https://github.com/Reading-eScience-Centre/coveragejson/issues/29#issuecomment-219984916

Maik: and how to define static subsets based on axis indices like the cube indices

not finished yet

<billroberts> Rob's wiki page: https://www.w3.org/2015/spatial/wiki/Data_cube_for_coverage

roba: joined OGC Europe on cisizen science observatories
... problem that cuts across SDW BP
... and a need to share resources
... developing a methodology for describing interoperability profiles, wanting to test and promote SDW BPs
... particularly want datacube descriptions to model large numbers of related datasets that realte spatially and temporally for citizen sciences to use
... want to test ideas coming from the group

<phila> Kerry: We seem to have forgotten the linking issue

<phila> ... It seems to have gone down in the list of things we need to deal with.

<phila> ... Are we making this up and we don't need it, or is it really critical - which I think it is

<phila> billroberts: I think it's important

<phila> Kerry: I'm not hearing it in the discussion today

<phila> billroberts: When I was working through the reqs, it does certainly come up in a lot of places

<phila> ... Metadata bout Coverage datasets, you need a mechanism to add metadata

<phila> roba: That's the basis of what we'll be demonstrating. using metadata to drive dynamic linking

<phila> ... Finding all the slices and dices through the coverage which we see as a linking use case

roba: want to use metadata to drive dynamic linking e.g. findin all slices and dices

<phila> Kerry: Sounds like a discovery use case to me

<phila> roba: That's a valid use of links

jonblower: but the melodies reasons are data annotation and linking to vocabularies that describe properties

billroberts: "omn the web" is important and sometimes its a simple as linked into a web page

<billroberts> https://www.w3.org/2015/spatial/wiki/Coverage_draft_requirements

Requirements doc: presentation and review of draft by Bill

billroberts: worked through UCR document and also BP document
... worked through those high level requirements\
... particularly coverage extracts in a web frinedly way
... some things are clearer and some are not
... q+
... will continue to tidy up and give me your thoughts...

and work through one issue at a time...

scribe: classification of requirements, started with coverage-related and some that should be
... plus ones we brought up in meetings so far...

billroberts: ... some things we might want to pick up on
... one thing is "discovery" e.g landsat archives
... not well solved at present for non-expert community
... another issues is identifiers for coverage extracts -- is such an identifier an aip spec for generating an extract
... or is it a thing?
... not possible to enumerate pre-prepared all possible exptracts, so you need some kind of api,

<jonblower> got a lot of half-baked thoughts on this that we probably don't have time to go into today...

billroberts: we also want to make sure we have a good connection to ssn strand of work

<jonblower> (I mean I have, I wasn't minuting Bill's thoughts!)

billroberts: seveal places we see that time is of equal status to spatial dimension
... some additional requirements not mentioned in UCR section
... also deciding what our target user groups are
... what is easy for some person is not so good for others
... planing to schedule a few of these per discussion

<phila> Kerry: I had a look at the doc and had some ideas for little bits here and there.

<phila> ... Can I suggest that you invite people to commnet on the doc in situ?

<phila> ... I think it's better that people edit the wiki and attach their names/comment

<phila> ... That might work better than e-mails with bits and pieces flying around.

<phila> billroberts: +1

billroberts: yes, please edit wiki directly, and use the "this is a minor edit" to not send emails if appropriate

<phila> Wiki pages have a history tab so you can see who has done what

billroberts: this is good for a frequent save strategy
... please stick you name on your comments
... BP meeting next
... any further comments?
... good stuff being done, thanks!

<billroberts> bye

Summary of Action Items

Summary of Resolutions

  1. approve minutes
[End of minutes]