Minutes from Best Practice deliverable group

SDW WG Face to Face, 11 March 2015

DRAFT

Participants

• Ed (Ed Parsons) - chair
• Frans (Frans Knibbe) - leading UC review
• Chaals (Charles McCathie Nevile) - leading UC review
• AndreaPerego (Andrea Perego)
• AndreasHart (Andreas Hart)
• DanBri (Dan Brickley)
• Dimitri (Dimitri Sarafinof)
• Gabriel (Gabriel Kepeklian)
• Kostis (Kostis Kyzirakos)
• Linda (Linda van den Brink)
• Paul (Paul van Genuchten)
• Phil (Phil Archer)
• Uwe (Uwe Voges)
• Xavier (Xavier Lopez)
•  ?? (?? ??)
• to be completed

Scribes

• Morning session (1): AndreaPerego
• Morning session (2): Ed
• Afternoon session: Linda

Morning session (1)

scribe: AndreaPerego

Ed

How we proceed?

Frans

Let's start with those marked with "BP"

Linda

We have one without the BP tag, but it's a BP one

Ed

Yes, it's mine

Paul

Can I suggest "spatial discovery" as a sub group

Ed

+1

UC-1: Style Template Use Case (Best Practice, Time, SSN, Coverage, Unsure)

rejected

UC-2: Meteorological Data Rescue Use Case (Best Practice, Time, SSN, Coverage)

Ed

there's the discovery element

Ed

let's keep it

UC-3: Habitat zone verification for designation of Marine Conservation Zones (Best Practice, Time, SSN, Coverage)

Xavier

It includes time

Ed

Let's keep it

UC-4: Real-time Wildfire Monitoring (SSN, Best Practice, Time)

Ed

(explaining the scoping questions)

Kostis

(explaining the use case, and the work done in TELEIOS)

discussion moves to assess criteria for UCs

AndreaPerego

(reading the BP deliverable description in the charter)

Ed

Can we use a 1-5 score

Discussion on URIs

Ed

So, about URIs,

Frans

Minting URIs for things is difficult in general, not only for the spatial domain

Ed

Should URI patterns represent a hierarchy reflecting the one of the things they're denoting?

Paul

(discussion about real world things, spatial object, and URIs)

... (URIs are not for the real world entity, but the representations of it)

AndreaPerego

outside the geo domain, this distinction is difficult to understand

Xavier

The purpose of BPs is to share with other domains expertise of the geo domain

Ed, Frans, Xavier

(discussion on hierarchical URIs, and persistent URIs)

Ed

Hierarchical URIs can be used to reflect hierarchies that are already in spatial objects

Ed

You can have also multiple hierarchies.

Ed

It would be good to propose a way to structure URIs for spatial things/objects.

Ed

One of the problems that currently we are not using URIs

Uwe

(pointing out the importance of context)

Paul

How these recommendations will impact at implementation level?

Xavier

URIs are a starting point to opening up geo data to a broader community

... the issue is also what we would like to denote with URIs

Uwe

In ISO 19115 what is described by metadata is defined by the "scope", that can be dataset, series, feature...

Ed

We should define the lower level of granularity

Ed, Frans

(discussion on location URIs, basic attributes, persistency and evolution over time)

AndreasHart

PIDs are a research issue that probably we should not address.

Ed, Xavier

(discussion on hierarchical URIs and spatial relationship)

Ed

We should consider adding a requirement about the need to design systems that are largely machine-to-machine

Frans

How should we phrase a requirement for URIs?

Paul

How we can use a URI in a WFS?

Ed

(summarising the URI issue for DanBri and Chaals, who just stepped in)

Ed

The issue is to use URIs to make geo data visible on the Web

Frans

Also with existing geo technology it may be possible to mint and maintain URIs in a scalable way - e.g., by re-using primary keys in a relational database.

Ed

Indeed. The system behind can be anything, it shouldn't be modified. The important thing is to have a URI, not how it is minted.

discussion on WFSs, and technological solutions to enable URIs

Coffee break - 10:50

Morning session (2)

scribe: Ed

AndreaPerego

Matrix of use cases versus Best Practice Reqs, e.g ontology, URI, Metadata, API’s

Ed

Best Practice should work at different scales, global, personal, microscopic.. and multiple reference frames

UC-8: Publishing geographical data (Best Practice)

Frans

Publishing geographical data use case - what requirements lead from this…

Xavier

Spatial data ? Only - Seems not

AndreaPerego

Related to apis ( restful)

Frans

Big question, software choice for example from the perspective of the publisher..

Dimitri

Many standards already exist, which ontology.. Don’t reinvent

Xavier

Multiple endpoints might help..

Chaals

Difficult to match ontology - translation is difficult

Frans

A “thing” is a geographical feature, from GeoSPARQL

AndreaPerego

Take best from different ontology

Frans

Make testable ?

DanBri

Can we not just compare between, definitions need not completely match.

Frans

Registry of reference systems ? - accessible via URI

Frans

Level of detail/Accuracy...should be published via Metadata - Important for machine to machine applications - errors multiply.

Chaals

Metadata should be a minimum set that is useful, so need to publish is a specified way.

Frans

Should we test software or provide tests for publishing tools

Xavier

First step to define interfaces, ontologies… etc validate from use cases

Chaals

Value of data proved by usage, Apple Maps users migrate to Google Maps for example :-)

UC-9: Consuming geographical data in a web application (Best Practice)

Frans

Consuming data from the perspective of web developer

Chaals

need to make content crawlable, then able to ask search engine or other service

AndreaPerego

Need to add RDF to geoportals to allow them to work with Search Engines. (see also UC-46)

Gabriel

Authority of wikipedia not proven…

Chaals

Provenance important, but we have mechanisms for identifying authoritative information already.. http://www.ushahidi.com/ good example

Frans

What data can I expect from a publisher ? 3D, Raster, etc

Chaals

Same as Data Shapes issues ? Falls out of Ontology, not geographically specific

Frans

Can I get the data with coordinates that match the coordinate system of my map?

Chaals

This is a core requirement, key geospatial thing we do - need common CRS and method to transform..

Xavier

Solved for Geo, WGS84 for 95% but explicit different ones for science for example.

Frans

Spatial extent, RDF version of bounding rectangle ?

Phil

DCAT can be extended for this.. Application profile under review.

Uwe

Dcat geo extension under development.

Chaals

Need a way to describe a spatial features as a shape to filter results.

Ed

Need both defined and fuzzy boundaries

Linda

Should be looking for things not datasets ?

Xavier

Search is one way to access data, other approach INSPIRE linked databases with SPARQL endpoints - Don’t know which is best. Will gov be willing to make data searchable - Interpretation of problem space linking metadata, not actual data.

Uwe

Model at lowest attribute level ?

Phil

Shapes not a good term without context

Lunch 13:00

Afternoon session

Back from lunch 14:45

scribe: Linda

Ed

does someone want to take us through the rest of the use cases?

Chaals

(volunteers)

Chaals

use case consuming geodata on the web, continued. question #8 is a really good question.

...A lot of data on the web is unclear. We need to make the easy things easy for 'dumb' users.

??

who is the audience for that question?

Chaals

a crappy developer who doesn't know geography. Seriously: an app developer who wants to do something spatial.

??

is it a goal to teach these audiences how to work with spatial data on the web?

Chaals

no, just make it possible. it's a requirement not to require quality data. But handling quality data is a req as well.

...the next question in this use case: what does the endpoint support? This is a general problem, not specific to spatial.

Frans

so we rely on the general solution coming out of the data shapes activity?

Chaals

yes, or out of the RDF architecture we are basing ourselves on, in general.

Clemens

are we saying spatial data on the web should be rdf?

Chaals

I kind of am.

next question

can i use spatial operators in federated queries? Short answer is yes, it's obvious. And federated querying is a general requirement, not specific to spatial. A req is: to be able to mix spatial operators with general ones in queries. A general req, not spatial specific, is to be able to ask an endpoint what questions it can answer.

next q

re responsiveness of my application: use caching.

Frans

the req could be that it should be possible for a system using spatial data on the web to be fast.

Chaals

this is a standard req, not specific to spatial.

Frans

spatial data can be very high volume. Geometries can be big strings.

??

is the q related to federated querying?

??

out of scope, move along

UC-10: Enabling publication, discovery and analysis of spatiotemporal data in the humanities (Best Practice, Time)

Chaals

(moves along to the spatial data in humanities use case)

??

looks like no new reqs in here. Unless: req named places can overlap and change over time.

ACTION: Chaals to write use case for fictional places

UC-11: Publishing geospatial reference data (Best Practice)

Chaals

publishing geospatial reference data. what is reference data?

Clemens

data that everyone understands to refer to something. req is to be able to annotate this data with a specification of what the information is / where do you find the geographic information for the wellknown reference like a zip code.

Ed

And this goes to the canonical source.

Chaals

i disagree. there is no one canonical source, it's different for different people.

Ed

in some cases there is, like the cadastre for parcel boundaries.

Chaals

there may be more than one authority sources for a bit of data.

Linda

it's a general thing again, lets move on.

Clemens

ok, but is another group resolving this?

Linda

provenance group?

Clemens

could also be http. where can we find the group to answer this. maybe the data on the web group.

ACTION: Clemens to look at data on the web draft to see if they address how to refer from data to the relevant authority sources.

Chaals

the req could be we must be able to match records.

AndreaPerego

what are we going to do with general reqs? we should keep track of them

Chaals

(agrees)

UC-12: Integration of governmental and utility data to enable smart grids (SSN, Best Practice)

(smart grids. merging of geographic with nongeographic data without fully understanding the complexity of each domain. No new reqs.)

UC-13: Using spatial data from the web in GIS systems during emergency response operations

(emergency response)

Chaals

req is to be able to use data coming from different systems, not only linked data. eg. wfs.

UC-14: Publication of air quality data aggregations (SSN, Time)

(quality data aggregation. skipped because not relevant to best practice.)

UC-15: Publication of transport card validation and recharging data (SSN, Best Practice)

(transport card validation. req is that places can be movable objects.)

UC-16: Combining spatial RDF data for integrated querying in a triplestore (Best Practice)

(combining spatial data for integrating in a triple store. copy/past the four bullets from this use case to the list of reqs.)

UC-17: Dutch Base Registry use case (Best Practice)

(dutch base registry. no new reqs.)

UC-19: Publishing Cultural Heritage Data (Best Practice, Time, Coverage)

(cultural heritage data. no new reqs)

Frans

maybe a req is to be able to validate data, specifically geometries.

Clemens

(agrees)

Chaals

however you should be able to use invalid data.

Kostis

we should not deal with the validation issue, but just try to consume data whether its valid or not.

Chaals

however we should the possibility to validate, to be able to discover if you will be able to use data.

Kostis

we should delegate this to the serialization formats we use.

Clemens

the bp should pick a serialization that has a validator, and pick a validator to use.

UC-20: Dissemination of 3D geological data (Best Practice, Coverage)

(3D data. req: 3D geometry / multidimensional data.)

Kostis

isn't this mostly not vector data but raster data?

Linda

there's also 3d vector data, eg citygml

Clemens

or geological boreholes

UC-21: Publication of Raw Subsurface Monitoring Data (Best Practice, SSN, Time)

(raw subsurface monitoring data. No new reqs.)

UC-22: Use of a place name ontology for geo-parsing text and geo-enabling searches (Best Practice, Time)

(geoparsing text. No new reqs.)

UC-23: Driving to work in the snow (SSN, Time, Coverage)

(driving to work in the snow. No new reqs,)

Frans

is the bit about discovering observation streams not new? Pull vs push?

Chaals

req: we need the data to be streamable.

Frans

sidenote: might be difficult. usually for coordinates in a format like wkt you need the whole thing up to the end before you can use it.

UC-24: Intelligent Transportation System (Best practices, Time, SSN, Coverage)

(intelligent transport systems. No new reqs.)

UC-25: Optimizing energy consumption, production, sales and purchases in Smart Grids

(optimized energy. Not discussed)

Ed

we have 10 minutes left and need to come up with one slide summarizing our progress.

Meeting closes

Summary of Action Items

[NEW] ACTION: Chaals to write use case for fictional places [yet to be added to the tracker]

[NEW] ACTION: Clemens to look at data on the web draft to see if they address how to refer from data to the relevant authority sources. [yet to be added to the tracker]

[End of minutes]