Re: [Minutes-Cov] 2016-03-09

Hi all,

Copying Maik as he’s not on the WG mailing list, but should hopefully be there soon.

Apologies for not being able to join the meeting today, as usual. This isn’t a great time for me in general, but if there is a timing change when the clocks change that might be better. But I appreciate that we need to find a time that works for all, which is difficult!

Anyway, thanks for the great minutes. I strongly support the idea of Kerry’s group (and indeed anyone else!) giving some independent scrutiny to CoverageJSON. This is the perfect time to do that. I also agree that some kind of comparison of the CovJSON model and QB would be really instructive. And that it’s entirely possible that CovJSON could use some terms from the QB vocabulary - potentially not just for the domain but also for the range metadata (which we call “parameters”). Overlaps with OGC DGGS would be great to explore as I don’t know much about this at all.

One question I have about QB is this: from my understanding it doesn’t have the concept of an ordered axis - the axes/dimensions are categorical in nature. Would we need to modify QB in order to inject these semantics? Almost all the axes/dimensions one would use for a coverage (latitude, longitude, time, depth, etc) have an inherent ordering, which is very important for applications.

Just to finish off, a small correction - CovJSON *does* support point clouds (and timeseries, and trajectories, and lots of other coverage types), but we just haven’t created examples for all these things yet. We should do this!

And just a final final point to clarify in case it’s needed - there is plenty of LD in CoverageJSON. Much of the metadata is bona fide JSON-LD, i.e. RDF. But we don’t try to put the verbose, array-oriented stuff into JSON-LD, it’s not a good fit.

Looking forward to some interesting work in this group!
Jon





> On 9 Mar 2016, at 21:04, Phil Archer <phila@w3.org> wrote:
> 
> And the final sub group meeting for this week, on Coverages, has its minutes at https://www.w3.org/2016/03/09-sdwcov-minutes.

> 
> Thanks to the team at ANU for being there and being ready to get stuck in experimenting with CoverageJSON, Data Cube and more. The action items are around refining the requirements coming from the use cases that we have.
> 
> A text snapshot is provided below as per usual.
> 
> 
>                        SDW Coverages Sub Group
> 
> 09 Mar 2016
> 
>   [2]Agenda
> 
>      [2] https://www.w3.org/2015/spatial/wiki/Meetings:Coverage-Telecon20160309

> 
>   See also: [3]IRC log
> 
>      [3] http://www.w3.org/2016/03/09-sdwcov-irc

> 
> Attendees
> 
>   Present
>          dmitrybrizhinev, billroberts, sam_t, PhilA, kerry, Duo,
>          Yadu
> 
>   Regrets
>          scott
> 
>   Chair
>          Bill
> 
>   Scribe
>          phila
> 
> Contents
> 
>     * [4]Topics
>         1. [5]Patent call
>     * [6]Summary of Action Items
>     * [7]Summary of Resolutions
>     __________________________________________________________
> 
>   issue-18?
> 
>   <trackbot> issue-18 -- Model Reuse -- open
> 
>   <trackbot> [8]http://www.w3.org/2015/spatial/track/issues/18

> 
>      [8] http://www.w3.org/2015/spatial/track/issues/18

> 
>   <billroberts>
>   [9]https://github.com/Reading-eScience-Centre/coveragejson/

> 
>      [9] https://github.com/Reading-eScience-Centre/coveragejson/

> 
>   <billroberts>
>   [10]https://github.com/Reading-eScience-Centre/coveragejson/blo

>   b/master/spec.md
> 
>     [10] https://github.com/Reading-eScience-Centre/coveragejson/blob/master/spec.md

> 
>   <scribe> scribe: phila
> 
>   <scribe> scribeNick: phila
> 
>   billroberts: No minutes of the previous meeting to worry about
> 
> Patent call
> 
>   <billroberts>
>   [11]https://www.w3.org/2015/spatial/wiki/Patent_Call

> 
>     [11] https://www.w3.org/2015/spatial/wiki/Patent_Call

> 
>   <kerry> [12]https://www.w3.org/2015/spatial/wiki/Patent_Call

> 
>     [12] https://www.w3.org/2015/spatial/wiki/Patent_Call

> 
>   billroberts: Main suggestion for today is to try to go over the
>   scope of the things we're meant to do, set objectives and then
>   get on with it.
>   ... Maybe a schedule may be too ambitious for today but we can
>   get some ideas.
> 
>   <billroberts> [13]https://www.w3.org/2015/spatial/charter

> 
>     [13] https://www.w3.org/2015/spatial/charter

> 
>   billroberts: Points to the charter's wording related to
>   coverages.
>   ... In Amersfoort it was clear that we're interested in all
>   sorts of coverages and subsets.
>   ... In the F2F meeting we starting to home in on the
>   CoverageJSON work from Uni Reading as a good starting point
>   ... We shoujld probabyl check that people think that's a good
>   idea.
>   ... And what our requiremetns are for a Recommendation - what
>   does it need to do to satisfy the WG, making it Webby enough.
>   ... So, comments welcome
> 
>   kerry: I'm afraid I wasn't in Amersfoort, and only saw the demo
>   from a distance. I would like to push a bit harder to get he
>   RDF data Cube working with what we do.
>   ... Which may be a query model rather than a delivery model.
>   ... I'm keen not to take Coverage JSON at face value, we should
>   try to go a little harder to go on the LD route.
>   ... There's something from Munster that had more of a data Cube
>   flavour to it, but it dealt with smaller sets
>   ... But I don't want to deny that what's being done at Reading
>   isn't very good.
> 
>   billroberts: I think Reading have attempted to come up with a
>   representation that works well with typicl Web tools but it's
>   not the only possible approach.
>   ... So some way of testing options against criteria would be
>   good.
>   ... is option A better than option B anad in what respect.
>   ... That might be a difficult thing to do. What makes something
>   Webby? We can link to it.
> 
>   <kerry> +q
> 
>   billroberts: Shoujld be able to request over HTTP. It would be
>   good if the bits of HTTP like conneg could be used.
>   ... There's an interesting discussion around conneg around max
>   resolutions, max data size etc.
>   ... And it needs to play nicely with other common Web
>   approaches.
>   ... That's what makes it suitable for being on the Web, but
>   that's not a very retsrictive set.
>   ... But UI'd be happy to have a strand that looks into the
>   CoverageJSON work and one that looks at QB, or an evolution of
>   it.
>   ... Especially when it comes to subsetting. I think QB has a
>   lot of potential on that.
>   ... QB is verbose, so if the coverage is small enough, that
>   won't matter.
> 
>   kerry: I agree. We do have a list of well developed ideas about
>   what makes data more webby in our own BP doc.
>   ... And I think we might be making a rod for our own back if we
>   try and deliver everything.
>   ... Maybve we just need to point to what we think is most
>   important, for a start. Good characteristics etc.
> 
>   ->
>   [14]https://github.com/Reading-eScience-Centre/coveragejson/blo

>   b/master/spec.md Example 1.1
> 
>     [14] https://github.com/Reading-eScience-Centre/coveragejson/blob/master/spec.md

> 
>   <billroberts> phila: describes what's in this example. In a few
>   lines of JSON, it sets out all the domain details
> 
>   <billroberts> phila: covers multidimensional grids, not (yet)
>   point clouds
> 
>   <billroberts> phila: that domain specification could
>   potentially be defined in RDF, either RDF Data Cube or
>   something that maps clearly to it
> 
>   <billroberts> phila: CSVW WG defined how to define metadata for
>   tabular data that defined how it could be transformed to RDF
> 
>   <billroberts> phila: some commonality there wtih the things we
>   have to do
> 
>   <billroberts> phila: Maik from Reading emphasises that
>   CoverageJSON is early experimental work
> 
>   <billroberts> phila: the opportunity to have the ANU team to
>   look at this in detail and suggest improvements
> 
>   <billroberts> phila: metadata is easy
> 
>   <billroberts> phila: providing values within the coverage prob
>   doesn't make sense as triples, which would be too verbose
> 
>   <billroberts> phila: but defining the domain in a way that
>   draws on data cube looks promising
> 
>   <billroberts> phila: this needs testing, I may well be talking
>   rubbish
> 
>   <billroberts> phila: it would be good to investigate if
>   CoverageJSON and RDF Data Cube can be combined in some way
> 
>   kerry: That sounds like a good plan. What Chris mentioned in
>   the last meeting...
>   ... the QB may not be granukar enough. WE need dicing as well
>   as slicing - which makes sense
>   ... And the other comment was... what do the team here think?
>   ... This is a student project and they get a lot of say in what
>   they do. It's a fairly open-ended spec
>   ... the spec is to work with the WG and deliver stuff that the
>   group can use
>   ... The group here is already tied to an infrastructure
>   environment that they're working with. It's different from what
>   the Reading is working on.
>   ... and they're working with another OGC spec DGGS
>   ... whichy seems to fit very well.
> 
>   Duo: Hi... I guess my role as team leader, client application
>   developer.
>   ... What we've been working on is very rought, just setting up
>   the raw data that we have and playing around with it.
>   ... So we haven't started looking at what we might use it for,
>   so we're open to what you'd like us to do with it.
> 
>   <billroberts> phila: the working group can interpret the
>   charter as it sees fit. We don't necessarily have to make a
>   specification
> 
>   <billroberts> ...there isn't a predefined answer, so it's ok to
>   follow the directions that come up in the research
> 
>   billroberts: The basic objective is clear - there is this big
>   collection of data from satellites etc. that is not being made
>   best use of in the Web community because it's too difficult
>   sand if we can make it play more nicely with the Web then
>   that's good.
>   ... One task that I had from a few weeks ago was to go through
>   the existing use cases to see how they look fromn the POV of
>   our coverage work.
> 
>   -> [15]https://www.w3.org/TR/sdw-ucr Use Case doc
> 
>     [15] https://www.w3.org/TR/sdw-ucr

> 
>   billroberts: What's harder to tell is whether they are
>   sufficient. I;m not sure that the translation from the UCs to
>   the Reqs is helpful enough to head towards a solution.
> 
>   action-144?
> 
>   <trackbot> action-144 -- Bill Roberts to Review the use cases
>   from a coverages point of view -- due 2016-02-16 -- OPEN
> 
>   <trackbot> [16]http://www.w3.org/2015/spatial/track/actions/144

> 
>     [16] http://www.w3.org/2015/spatial/track/actions/144

> 
>   billroberts: I can document what I came up with and make some
>   suggestions for refining those requirements. That would be a
>   starting point, and we have the BPs as a guide too.
> 
>   phila: Will there be a new version of the UCR?
> 
>   kerry: Yes. And Frans is expecting the sub groups to come up
>   with better stuff that needs to be incorporated.
>   ... A more refined description of the requirements.
> 
>   billroberts: The reqs at the moment are very much at the
>   summary level. I think we can probably fill those out in a bit
>   more detail.
>   ... A suggestion - could the ANU team take a look at the UCR
>   and see how that fits with the AGDC data Cube data
>   ... What kind of applications do you see? etc.
> 
>   Duo: I think that at this stage, we are still trying to figure
>   out what you guys are doing. We're also working with Kerry and
>   Matt from AGDC who might be having different requirements from
>   the SDW WG.
>   ... We need to work with both teams.
>   ... But it would be good to get clarity on what you want.
> 
>   kerry: Just to support that. The team is not use case driven.
>   They're not working with users of the data. Matt does work with
>   data users though.
>   ... Matt is looking for something that in his work with GDDS
>   can deliver against the interoperability requiremetn that he
>   has. He's not imposing abything beyond what this group is
>   doing, just to be able to use GDDS.
>   ... It really is open ended. Taking the advice from this group
>   is the way to get maximum impact from the project.
>   ... The team is comfortable with it being very experimental.
>   ... They're well matched for it to be an experimental project.
> 
>   billroberts: Something I'd find useful, is some good worked
>   examples of you taking some example satellite data of interest
>   and showing what are the steps you have to do to get hold of
>   it, process it, filter it, how it refers to POIs on the ground
>   etc.
> 
>   <billroberts> phila: agrees. Running code beats everything.
> 
>   <billroberts> phila: the Reading work has been a compelling
>   example because it clearly works and we can see that it takes
>   satellite data and makes it usable in a browser
> 
>   <billroberts> phila: would be great to see similar things from
>   the ANU team
> 
>   <billroberts> phila: this project is an opportunity for ANU to
>   demonstrate a lot of impact from their work through involvement
>   in the group
> 
>   Duo: I think everyone in the team is super keen. Everyone's
>   inetersted, especially as it is so open ended.
> 
>   <kerry> +1 to wiki
> 
>   billroberts: I propose that we spend time on that stuff around
>   use cases and requirements. Looking at the BPs in the context
>   ... I'll do that on the wiki in the next couple of days.
>   ... We can use the mailing list to prod people once that's in
>   existence.
>   ... From there then I guess we want to scope out possible
>   solutions and check them against the requirements. Maybe Maik
>   can do that.
>   ... Maybe he can look through that in the context of the
>   CoverageJSON stuff, might he need to/want to tune his work etc.
>   ... Would you, Kerry, be interested in how the Data Cube might
>   relate to that?
>   ... Exploring and making some working notes?
> 
>   kerry: I'd live to say yes but honestly I'm not sure I can
>   between now and the next meeting.
>   ... (I have SSN stuff to work on etc.)
> 
>   billroberts: I;m happy to do a little on that. I'll at least
>   make a QB for Coverages wiki page and see if we can promot
>   ideas from other people
> 
>   kerry: I'm intrested and keen on this but I'm not sure how much
>   I can add given other roles in the WG.
> 
>   billroberts: A procedural thing...
>   ... Our first milestone is the FPWD. WE have to work out a
>   draft of what?
>   ... I think I've got what I was hoping to get today...
> 
>   kerry: A question about turnout - is this going to be it for
>   the regulat bi weekly meetings?
> 
>   billroberts: I don't know. Other people have expressed an
>   interest... Scott was keen. Of course this week is OGC TC week.
>   Robin was interested.
>   ... Peter Baumann said he definitely wanted to take part.
>   ... The timing of the calls rules out Chinese folks.
>   ... And Maik Riechert we can hope will join the call.
> 
>   kerry: So we might get a few more people/
>   ... But we do need to have a few more active people in the
>   group.
>   ... If it's just us, we can possibly come up with a better time
>   (for Aus and UK). This time is good for the USA as well
> 
>   billroberts: If they do join then great. If not then we may
>   look again, especially when the clocks change
> 
> Summary of Action Items
> 
> Summary of Resolutions
> 
>   [End of minutes]
>     __________________________________________________________
> 

Received on Thursday, 10 March 2016 06:42:04 UTC