Spatial Data on the Web SSN Sub Group Teleconference -- 11 Oct 2016

<ahaller2> trackbot, start meeting

<trackbot> Meeting: Spatial Data on the Web Working Group Teleconference

<trackbot> Date: 11 October 2016

<maxime> what is the webex pwd ?

<SimonCox> pwd at the top of this screen

approve minutes

<ahaller2> https://www.w3.org/2016/09/27-sdwssn-minutes

<kerrry> +1

<DanhLePhuoc> +1

<ClausStadler> +1

<maxime> +1

<SimonCox> +1

<kerrry> patent call: https://www.w3.org/2015/spatial/wiki/Patent_Call

<KJanowic> present Kjanowic

<ahaller2> scribe: kerry

<kerrry> scribe: kerry

<ahaller2> scribenick: kerry

<kerrry> scribenick: kerry

redirection of urls

Demo of Slash URI ontology submodule implementation

<kerrry> maxime show presentation

<kerrry> maxime: for smart energy aware systems

<kerrry> ...required the w3id namespace to be the only one for all the ontology mpodules and versions

<kerrry> ...i will show how

<ahaller2> http://ci.emse.fr/seas/

<kerrry> ...browser does not show rdf+xml

maxime: if i go to this uri [shows]

[scribe had trouble missed a lot]

maxime: this is not what was requested put there is info as a cooluri to redirect to the vocab that you are loooking for
... so the latest is how to deal with alignments
... the main seas ontology and one module with a specific version
... the documenation says ..note SDW bp 11
... this one is the pep ontology
... says align with ssn
... the lates version of pep is this doc in html, this doc in ttl,
... thist is an alignment with ssn
... and I get the html by defualt and I can also get the ttl

this alignment imports pep and ssn

maxime: this is how we deal with external alignments...
... import the ontologies that are alsigned...

<KJanowic> this is also how we do it in GeoLink

armin: questions?
... to clarify: all the differnt turtles are separate files?

maxime: yes they are different

armin: so this would allow us to have differnt files for differrnt modules with /uris to point to different moduels to differnt files

<KJanowic> armin: so modules are still on the table, right?

rob: 3 questions

roba: when u redirect to a file why not the #identified in the file?

maxime: yes, we well have to do this
... if i got to any there is no # and I just target the top of the html element

roba: # based urls, not uris
... if we look at ontologies of sensor types they will not be in one place... what is the implication for large ontologies?
... can the nodes of a federation do something like this?

maxime: do you mean namespaces with multiple hosts?

roba: what is mimimum for compatibility
... the nevery host will need its own implementation choices
... what behvaious is epcified?

maxime: "cooluris" does this

<ahaller2> https://www.w3.org/TR/cooluris/

roba: is this necessary and sufficient?

maxime: we are working on some lists of classes of sensor
... if someone wants to create some new resouces e.g. new subclass of a sensor

roba: has an alternative model with demonstrate

<ahaller2> kerry: the old ssn follows the cool URI approach. What were you referring to?

<ClausStadler> the issue was the content-disposition - so that you get a nice download name

<ahaller2> maxime: Old SSN does not have any modules.

<ahaller2> maxime: if I go to the URI of SSN I don't get what I want

<SimonCox> I don't want to take any more time on this tech issue in this call, but would like to ask (1.) is Content-Disposition a new HTTP header (2.) when you say the property URI (a slash URI I think) is "not on the web" what do you mean? if it resolves to something on the web, then it looks like it is indeed "on the web"

<ClausStadler> e.g. the browser could offer for a file called Ontology.rdf using content-disposition a nicer filename such as SSN-Ontology-1.0.rdf

<SimonCox> Lovely demo , btw

armin: an idea now on how we want to implement

https://www.w3.org/2015/spatial/wiki/SSN_Tasks#Implementation_requirements_for_Rec-Track_document

armin: follw up on email list
... asking for commtments to implementations of new and changed terms
... we have 3 options
... we can have modules and extensions in rec track do as non-nornative or in a separate note or in ful rec track normative

<SimonCox> What does 'implementation' mean?

armin: as long as we have a redirect from old terms to new terms we can use the old implementations

<maxime> +1

armin: e.g. phenomenon time that is not in ssn

<maxime> as we have equivalence relations with the old ssn concepts

armin: no qequivalent relation, we would need to show an implelentation
... who would show this?
... please say if you can

<SimonCox> I can do some work

[some yeses]

<KJanowic> Could we clarify a bit more what this would take?

<KJanowic> I can do some work

SimonCox: what does implementation mean?

<KJanowic> It means prior use, right?

<DanhLePhuoc> +1 for the Simon question

<KJanowic> I think this is about semantics. there is the 'implementation' in terms of coding the SSN/SOSA and the 'implementation' in terms of showing prior use

<ahaller2> roba: if inferences in an existing implementation still work after changes, that I would see as evidence

<KJanowic> Unfortunately, this cannot be shown in general but only by specific use cases

https://lists.w3.org/Archives/Public/public-sdw-wg/2016Oct/0031.html

<KJanowic> no, impossible

armin... this would change to the new ssn and show that all the old classes still work

<ahaller2> maxime: agree as long as inferences work in existing implementation.

maxime: one would just need to change the owl:import axiom, and any inference that was possible before would still be possible thanks to the equivalence relations

KJanowic: 2 things, the term implementation has 2 flavours
... 1 is the actual coding
... 2 is showing in prior use
... these are confused.
... when it comes to entailment we cannot show this execpet for some use cases
... we can not do it in general becuase if we remove 500 classes as in dul we cannot show this

phila: apologies for voice problems
... so I hear my name about what implementation means

<SimonCox> In ORG document there is an Annex which tabulates a number of implementations.

phila: dcat and org ontology and rdf datacube have implementation reports

<KJanowic> So, prior use, right?

<SimonCox> e.g. https://www.w3.org/2011/gld/wiki/ORG_Implementations

phila: show you term by term whether an app has used this term
... if u report you used it we will beleive it...can you find running software that makes some use of each term
... in at least 2 pieces of independent software
... underlying idea is do you have classes that are genuinely useufl]
... evidence that you are doing somethinh useful is the big thing
... given that you have some legacy, if the old namespace works then we are ok
... you can propbaly adduce that evidence into the argument

<KJanowic> So old SSN and O7m count as pior work, i.e., implementation, right?

armin: that pretty much covers the case for the unchanged terms

<KJanowic> s/07m/O&M

armin: anyone would just need to cvhange the import to the new namespace

<ahaller2> kerry: has capacity march next year

armin: want to know who has the capacity to implement ?

KJanowic: this depends on whether we have a problem ..."demonstrating prior use" by use somewhere else before

kerry: no! clearly not.

KJanowic: so this is easy eg. using o&M implementations

phila: purl.org is up and running so you can redirect and use the old namespace -- that will work

phil: are u suggesting that lots of use cases for a partiular term will be taken as use cases? that will not work, must be a term in the software

KJanowic: so it can't be a dataset?

phila: yes data is fine too!

<ahaller2> ahaller: it can be a dataset, but it must use the URI that we define

<KJanowic> this is fantastic news

maxime: question: phil is it possible to change the redicrection of the old namespace to the new one

phila: yes, purl.org has been restored and you can redirect. we only need to redirect that namespace to the w3.org/ns namespace with a note that explains the redicrection
... was not possible 2 weeks ago
... with the caveat that I am not certain due to mobile phone, yes

phila: the bits of software need to be independent -- by 2 separate organisations.
... what does happen is a telephne call with chairs editors and director and the director makes a judgement
... if a term is implemented only once it would be looked at sceptically, it would be helpful if there are others
... you need 2 independet implementations
... one term with only one im might be ok, but not half the ontology

<DanhLePhuoc> +q

<ahaller2> kerry: proposes a survey of old implementations and the terms

<KJanowic> +1 to kerry, this cannot be done

<ahaller2> kerry: KJanowic thinks that terms that are implemented in O&M spec are sufficient, it is not ok, it has to use the namespace

<KJanowic> hmmm

<KJanowic> armin, I would just need 30 sec

ahaller back to who can implelement?

<KJanowic> I commit to implement some of that work

ahaller: I can do soemethin with acorn-sat in collaboration with the BoM

KJanowic: back to o&M stuff if we ppropose new terms then we do not have a namesapce
... software I have wriiten is ok right/

phila: maybe... a bit dodgy.... even better is someeone who is not a member of the grop

KJanowic: say the most basic "sensor" we only have to show there is some real software the namespace does not matter

<maxime> proposed structure for the wiki page that lists the existing implementattions of ssn:

<maxime> a) softwares that consume ssn terms

phila: no. you need to use the namespace. you cannot point to someon's iot sustem that has an "actuator' with no connection to ssn

<maxime> b) softwares that produce ssn terms

<maxime> c) datasets that use ssn terms

phila: you put the term in now and you ask for implementations in CR phasee
... you just have to prove that the new features work

<ahaller2> DanhLePhuoc: has capacity to implement some terms

DanhLePhuoc: can also implement , but do not yet know the new list!

SimonCox: is now confused

<KJanowic> Happy to implement as well

SimonCox: if the old ssn is prereq then what is staus of old ssn?

phila: current ontology is result of community group
... process of working group us taking this to full status
... if any existing terms are deprecated but not deleted ... we will confer formal status

SimonCox: so changing the semantics in subtle ways ?

<KJanowic> +1

phila: caution against subtle changes --- if you change the meaning do you break stuff? If you break stuff then be very careful

<SimonCox> I have to leave now ...

armin: summarises

<KJanowic> i can implement the SOSA core part

armin: please look at table as we will need implementations of sos core or will need to be non-normative
... no time for 3rd agenda item.

<KJanowic> bye bye

armin: thanks to all!

bye!

- DRAFT -

Spatial Data on the Web SSN Sub Group Teleconference

11 Oct 2016

Attendees

Contents

approve minutes

redirection of urls

Demo of Slash URI ontology submodule implementation

https://www.w3.org/2015/spatial/wiki/SSN_Tasks#Implementation_requirements_for_Rec-Track_document

Summary of Action Items

Summary of Resolutions

Scribe.perl diagnostic output