W3C home > Mailing lists > Public > public-sdw-wg@w3.org > August 2016

RE: UCR issue 30: missing requirement

From: Linda van den Brink <l.vandenbrink@geonovum.nl>
Date: Fri, 19 Aug 2016 06:14:52 +0000
To: Rob Atkinson <rob@metalinkage.com.au>, Frans Knibbe <frans.knibbe@geodan.nl>
CC: Joshua Lieberman <jlieberman@tumblingwalls.com>, Jeremy Tandy <jeremy.tandy@gmail.com>, SDW WG Public List <public-sdw-wg@w3.org>, "Payam Barnaghi" <payam.barnaghi@gmail.com>
Message-ID: <13F9BF0BE056DA42BFE5AA6E476CDEFE725FC1BF@GNMSRV01.gnm.local>
Is your hydrology example a case for BP8 (How to describe relative positions)? That would be interesting because we have an issue there: ISSUE 121 [1]
“Do we need this as a best practice; if yes, this BP needs more content”

[1] https://github.com/w3c/sdw/issues/121

Van: Rob Atkinson [mailto:rob@metalinkage.com.au]
Verzonden: vrijdag 19 augustus 2016 03:59
Aan: Frans Knibbe; Rob Atkinson
CC: Joshua Lieberman; Jeremy Tandy; Linda van den Brink; SDW WG Public List; Payam Barnaghi
Onderwerp: Re: UCR issue 30: missing requirement

Mainly it was looking ahead :-)  But IMHO it is important to get the intent, then wording, of such requirements right - is it for there to be guidance for how a community might solve such a problem, or is it for interoperability in the broader ecosystem of tools - i.e. the community is the virtual community of  W3C or OGC standards implementers.

GeoSPARQL is the latter case,
CRS definitions is the former - but one where the OGC makes recommendations and uses specific CRS definitions as defaults in some specifications.

The key thing for this requirement is whether vague descriptions are
a) purely textual annotations (in which case we probably dont need to say anything about them at all in the BP)
b) qualifications for a geometry property (in which case we probably want to define a vocabulary to identify such properties, and how to bind these to multiple geometries in a single feature - perhaps annotations need to apply to all provided geometries)
c) machine-readable constructs (possibly with qualifications) - i.e. the ability to say A isNear B

I would suggest its necessary for a BP to handle machine readable location descriptions with human readable annotations, i.e. cases b,c where A relatedTo B  and either A or B is a spatialThing. Note this covers providing a note about geometry per feature.

Thus, I'd be tempted to say - in the BP - if the relationship can be expressed using the GeoSPARQL specification, then this should be used, either directly or by specialisation to introduce domain specific semantics  domain-independent spatial operation. Otherwise follow the general BP regarding vocabulary re-use.

In, for example hydrology, a description of where a stream gauge is located relative to a stream confluence is actually far more precise than a coordinate somewhere near the confluence - which may be upstream, downstream or at the actual confluence in fact.

In the Requirements, therefore, I'd be tempted to rephrase "vague,imprecise"  to "non-coordinate based" - and then identify the above cases and state which is in scope.


On Thu, 18 Aug 2016 at 20:37 Frans Knibbe <frans.knibbe@geodan.nl<mailto:frans.knibbe@geodan.nl>> wrote:
Hello Rob,

Was your comment intended as criticism of the proposed rephrasing of the spatial vagueness requirement? Or is it only looking ahead to possibilities of meeting such a requirement?

Although the primary concern of this thread is to formulate the right requirement(s), I must admit in this particular case it is interesting to think of possible ways of making it possible.

Again I think a spatial ontology could be really helpful. Let's take some examples of text that might be turned into spatial data:

  1.  The Carthaginian army was defeated near the southwest border of the Roman empire.
  2.  The suspect moved from the entrance of the bank to a red car that was parked near the post box.
The first example might come from a historical source and the second could be an example of crowd sourced data, two use cases in which vague spatial data are typically encountered.

A hypothetical spatial ontology will have definitions of the concepts of 'spatial thing' and 'spatial relationship'. So at least we could flag the locations and the spatial relationships in these statements as such. That could already be helpful.

Now in the first example it is imaginable that a resource exists that defines the terms used in a domain context. Historians studying the Roman empire could make a vocabulary in which the general classes for location and spatial relationships are specialised. It could have a collection of linestrings marking the borders of the empire through time, and it could have a definition of the spatial relationship 'near', which in historical Roman texts could always mean 'a distance of at most one day's marching'. Furthermore, the spot where the battle took place could be represented as a 2D point geometry with unknown coordinates (by the way, a possible example of why the coordinates should not be a mandatory part of a geometry).

For crowd sourced information, definitions of vague terms that are used will probably be more difficult to outsource to domain vocabularies. Definitions of terms can be as diverse as the crowds using the terms. But at least flagging the locations and spatial relationships using the general spatial ontology could be useful.


On 18 August 2016 at 01:10, Rob Atkinson <rob@metalinkage.com.au<mailto:rob@metalinkage.com.au>> wrote:
IMHO this is covered by the general vocabulary re-use clause - such vague terms are domain specific semantics - therefore in general you should look to re-use a set of relationship properties, as defined in an ontology, published by the community of practice you intend your data to be understood by/interoperable with.

In general, one should look first  to the OGC for spatial concerns, to see if such a community has either published what you need, or has a governance structure in place (a Domain Working Group) where such a vocabulary can be shared. (note that OGC will reference relevant vocabularies published by other SDOs.... so its a sensible starting point IMHO)


On Wed, 17 Aug 2016 at 23:10 Joshua Lieberman <jlieberman@tumblingwalls.com<mailto:jlieberman@tumblingwalls.com>> wrote:
Can we distinguish between qualitative relationships such "bottom of the hillside” which are as precise as the features being referenced,  and fuzzy ones such as “near the hillside” that explicitly use imprecise relationships?


On Aug 17, 2016, at 9:00 AM, Frans Knibbe <frans.knibbe@geodan.nl<mailto:frans.knibbe@geodan.nl>> wrote:

Dear group members, especially the BP editors,

It would be great if we can resolve this sleeping issue<https://www.w3.org/2015/spatial/track/issues/30> before the next PWD of the UC&R document. To summarise the issue, it seems clear what the requirement is: there is a need to be able to use vague/informal/colloquial expressions to refer to either spatial things or spatial relationships.

I still think the easiest solution is to change the existing Spatial vagueness<http://w3c.github.io/sdw/UseCases/SDWUseCasesAndRequirements.html#SpatialVagueness> requirement a bit. The core requirement would then be something like "It should be possible to use vague or informal expressions to indicate locations or spatial relationships". That requirement could be followed by some examples:

for locations:

  *   at the bottom of the hillside
  *   downtown Los Angeles
  *   London (has multiple definitions, so just the name is not precise)
  *   the south west boundary of the Roman Empire
for spatial relationships:

  *   near
  *   across the street from
  *   upstairs
  *   at walking distance from
What do you think?


On 20 October 2015 at 14:04, Frans Knibbe <frans.knibbe@geodan.nl<mailto:frans.knibbe@geodan.nl>> wrote:

2015-10-16 11:15 GMT+02:00 Jeremy Tandy <jeremy.tandy@gmail.com<mailto:jeremy.tandy@gmail.com>>:
Hi Frans-

I'm not sure that your option (1) covers the terms used for 'vague' (or, more accurately, _relative_) spatial relationships. I think that we might want to refer to the location of a post box unambiguously, based on it's position within a topological (road) network; e.g. 150 from the junction of roads A and B in the direction of [etc.] ... the junction (a node in the network) might have a geometric position (e.g. collected by a surveyor with GPS), but the position of street furniture may be recorded using relative positions.

We already have a requirement for being able to use spatial relationships, see the Spatial relationships requirement<http://w3c.github.io/sdw/UseCases/SDWUseCasesAndRequirements.html#SpatialRelationships>. If that requirement is met, it should be possible to express the location of a post box relative to some topographic or topological point, wouldn't you say?

However, the ability to be vague about relative positioning does not seem to have been addressed yet. One might want to say that a post box is close to the butcher shop, or over the road from the bakery.


Does that help?


On Wed, 14 Oct 2015 at 13:17 Frans Knibbe <frans.knibbe@geodan.nl<mailto:frans.knibbe@geodan.nl>> wrote:
Rachel and Jeremy, thank you for helping us solve this case.

So this is about being able to use colloquial terms for both location and spatial relationships. It seems to me that the first part, colloquial terms for location is basically covered by the Spatial vagueness requirement<http://w3c.github.io/sdw/UseCases/SDWUseCasesAndRequirements.html#SpatialVagueness>. Interestingly enough, this requirement has not been related to the Best Practices requirement.

What we could do is:

  1.  Rephrase the spatial vagueness requirement a bit to make it clearly cover examples like “the midlands”, “town centre”, how different people define “London”.
  2.  Relate the spatial vagueness requirement to the Locating a Thing use case<http://w3c.github.io/sdw/UseCases/SDWUseCasesAndRequirements.html#LocatingAThing> and the Best Practices deliverable
For the requirement to be able to use colloquial terms for spatial relationships we could either expand the definition of the Spatial vagueness requirement, or add a new requirement, so that we end up with separate requirements for spatial vagueness for locations and spatial vagueness for relationships. I would favour the first option, to keep things simple, and because there is of plenty of overlap between the requirements.


2015-10-13 18:03 GMT+02:00 Jeremy Tandy <jeremy.tandy@gmail.com<mailto:jeremy.tandy@gmail.com>>:

Rachel is correct; 'Locating a thing' [1] (provided by @eparsons) is the source of this requirement. The description provided in her message is accurate. Ed also uses phrases like "upstairs", "where I left it" etc.

It's not about geocoding; it's about relating position in human terms ... all about context.

FWIW, there are already some reasonable models from OGC about describing relative positioning - usually related to position within a topological network offset from a node in that network (e.g. position of signage on a railway, position of a lamp post on a street etc.)


[1]: http://w3c.github.io/sdw/UseCases/SDWUseCasesAndRequirements.html#LocatingAThing

On Fri, 9 Oct 2015 at 17:37 Heaven, Rachel E. <reh@bgs.ac.uk<mailto:reh@bgs.ac.uk>> wrote:
Hi Frans

Looks like this is from the “Locating a thing” use case,

It’s about vernacular geography :  human terms for relative spatial positioning (“upstairs”, “over the road from”) and human concepts of places (“the midlands”, “town centre”, how different people define “London”). These extents are usually vague and do not match official authoritative boundaries, so you can’t geocode them accurately, if at all.

It will also be very relevant to harvesting crowd sourced data https://www.w3.org/2015/spatial/wiki/Working_Use_Cases#Crowd_sourced_earthquake_observation_information_.28Best_Practice.2CSSN.29


From: Frans Knibbe [mailto:frans.knibbe@geodan.nl<mailto:frans.knibbe@geodan.nl>]
Sent: 09 October 2015 14:11
To: SDW WG Public List; Kerry Taylor; Jeremy Tandy
Subject: UCR issue 30: missing requirement


This is the thread for discussion of UCR issue 30<http://www.w3.org/2015/spatial/track/issues/30>, the Case of the Mysterious Missing Requirement.

The current description reads: 'see " relative (spatial) relationships based on context e.g. my location [expressing location and places in human terms] " from
https://www.w3.org/2015/spatial/wiki/BP_Consolidated_Narratives#linking_data'. Jeremy might know what use case it came from.'

To me is not exactly clear yet what the requirement could be. Resolving location names in human terms to geometry is called geocoding and is a well established practice. Could this be about the need for having human language equivalents for spatial relations? I can see that would be a benefit for finding spatial data using a search engine.

If we find the related use case(s) we will probably get a better idea of what the missing requirement could look like,


This message (and any attachments) is for the recipient only. NERC is subject to the Freedom of Information Act 2000 and the contents of this email and any reply you make may be disclosed by NERC unless it is exempt from release under the Act. Any material supplied to NERC may be stored in an electronic records management system.

Received on Friday, 19 August 2016 06:15:35 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:31:25 UTC