Spatial Data on the Web SSN Sub Group Teleconference

31 Jan 2017


See also: IRC log


roba, ahaller2, Kjanowic, SimonCox, ScottSimmons, kerry, phila, Joshlieberman, Francois, DanhLePhuoc, RaulGarciaCastro, ClausStadler, sefkikolozali, Kerry (IRC only)


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

SOSA/SSN vertical integration architecture https://www.w3.org/2015/spatial/wiki/Proposals_for_rewriting_SSN#Compromise_Proposal_6_made_by_Kerry_January_2017

armin: please let us know if you have any copyright infringement.

Armin: we are going to go through the Kerry's proposal.

<Kjanowic_> and this is problematic

<Kjanowic_> +1

<roba> +1

armin: everyone agrees we have two uri two ontology files

<SimonCox> +1

<DanhLePhuoc> +1

<kerry> objection!

<kerry> I cannot hear the discussion as you know

<phila> PROPOSED: That SOSA and SSN have two different namespaces

<kerry> and i do not know anything about a resolution that just appeared on the irc

<Kjanowic_> +1

<SimonCox> +1

<kerry> I canot agree

<kerry> -1

<phila> +1

roba: two different namespaces are not the same as two different URIs

<Kjanowic_> We have "he namespace for SSN terms is http://www.w3.org/ns/ssn/" and "The namespace for SOSA terms is http://www.w3.org/ns/sosa/"

<SimonCox> Can we use the queue please?

<roba> two ontologies IRIs may reusee the same namespace (which has a RI which may or may not be an ontology IRI as well)

<phila> +1 to SimonCox

<Kjanowic_> I agree with SimonCox

<Joshlieberman> Simon, did you just ordain us?

simon: the comment that rob and josh has been making is strictly correct. The URIs and namespaces are strictly independent from each other. We should take into account the users for namespaces and URI decision

<roba> i agree also - i think this is the key issue

<SimonCox> <Kjanowic_> We have "he namespace for SSN terms is http://www.w3.org/ns/ssn/" and "The namespace for SOSA terms is http://www.w3.org/ns/sosa/"

<roba> are we voting on the state of the draft or the principle?

<Kjanowic_> the principle

<ahaller2> PROPOSED: That SOSA and SSN have two different IRIs and two different ontology namespaces

<Kjanowic_> +1

<SimonCox> +1

<phila> +1

<DanhLePhuoc> +1

<ahaller2> +1

<RaulGarciaCastro> +1

<kerry> -1

roba: whether ssn itself needs to have a different namespace or a separate ontology file is a separate issue. this issue is about files: packaging.

<Kjanowic_> I would have to disagree

<ahaller2> PROPOSED: That SOSA and SSN have two different IRIs

<roba> +1

<SimonCox> New issue: do terms defined in the SOSA ontology have a different namespace than the terms defined in the SSN namespace

roba: can we separate the questions to having to different ontologies with two different IRIs.

<kerry> -1

<ahaller2> kerry on wiki: Core ssn and extended ssn are presented as different ontologies - in dfferent files with different ontology iris

<SimonCox> Correction: roba asked do terms defined in the SOSA ontology have a different namespace than the terms defined in the SSN ontology

<SimonCox> Kjanowic_: asks if SSN will import SOSA and re-use the SOSA URIs where they overlap with the SSN concepts.

<Kjanowic_> no, we do

<SimonCox> kerry: no-one is proposing ssn:Sensor rdfs:subclassof sosa:Sensor - please look at the work I posted in the last week.

<roba> agree with jano - and armin - these are follow-on issues that we can address once principles are ticked off one by one

<SimonCox> Joshlieberman: rdfs:isDefinedBy is not the membership predicate for ontologies

<SimonCox> it is a common convention but not universal and not an axiom

<Kjanowic_> [but we aready voted on that]

<phila> Joshlieberman just said what I wanted to say only more clearly than I would have.

roba: we can use the ontology membership for classes, sosa classes belong to sosa and ssn classes. however, restriction are difficult to make for certain ontology. they are only group of a file. therefore, as a hint having different namespace c could help to separate. we still have to be careful so that when we make separation.

<roba> agree with Josh - we may need three files: SOSA, SOSA formal axioms and SSN refinements

<Kjanowic_> Yes, what does this do to our previous vote?

<Zakim> kerry, you wanted to say that these issues cannot be determinedin isolation. If this forces us into a position where we have to say that ssn:Sensor rdfs:subclassof sosa:Sensor then

<Joshlieberman> 1). Two ontology IRI's? - yes. 2) two namespaces? Maybe. 3) how to segregate axioms? Not sure

<kerry> I cannot hear a thing and very little is going into irc, I am afraid.

armin: we have two different files and two IRIs. we could separate the question.

<kerry> soory -- am i acked or not?

armin is reading Kerry's comments.

<kerry> I wanted to say that these are not issues that can be determined in isolation

<kerry> It looks a lot like boiling frogs to me

<Kjanowic_> I have to disagree

<kerry> for example, a combination of bits and peices in this issue have ssn (ontology) importing sosa (ontlogy)

<kerry> and uing sosa terms (namespace) and havinh ssn (ontology)

<kerry> having to include an axiom like ssn:Senor rdfsSubcallsof sosa:Sensor

<kerry> this makes ssn unusable and simply stupid.

<Kjanowic_> +1 to simonCox

<Kjanowic_> yes and no

<roba> agree - simon has show we dont need subclasses (+1_ - so the issue is naming preferences for SSN narrower cases ?

<phila> What Simon said made sense to me

Simon: my preference is to keep sosa as lightweight, and have a separate IRI.

<Kjanowic_> not really

<phila> ahaller2: Having separate files and namespaces doesn't itself imply subclassing etc.

<DanhLePhuoc> Simon's summary: https://github.com/w3c/sdw/pull/517

<roba> i thought we refined the proposal

<roba> so we didnt conflate namespaces

<roba> this is not a problem if SSN is providing axioms based on SOSA definitions - not "extending"

<SimonCox> +1 roba

<Zakim> phila, you wanted to talk about DC, DCAT and DCAT-AP

<roba> armin - not "two" - "separate" - maybe we'll decide on three....

<ahaller2> roba: yes, sorry, not "two", "different"

<Joshlieberman> Trying to separate 3 issues. Third is how to segregate SSN axiom from Sosa vocabulary without subclasses. Only different files and or separate namespaces can suggest that the axioms should not be applied to Sosa. No part of OWL enforces that.

<Kjanowic_> yes, exactly +1 to what phila is saying

<roba> yes + 1 = SOSA, SOSA OWL axioms _ SSN aplication profile is exaclty the pattern I was envisoning in the original vertical mod discussion

<Kjanowic_> +1 (YES!)

<Zakim> kerry, you wanted to say i have no idea what simon siad but that I saw this stement in simon's work on git hunb about 24 hours ago which is the last time I looked and to sak whther

<Kjanowic_> This is due to the semantics of owl, this is not about what we like but what the W3C sepcs state. So yes +1 to phila

<Kjanowic_> +1 to SimonCox

<Kjanowic_> IMHO 3 files are really overdoing it. I see the idiea but this will be too complicated

simon: two possible patterns to be used in ssn and sosa. one is subclass, other one is adding axiomization.and constraints to sosa.

<ScottSimmons> * sorry, must leave early

<Kjanowic_> Yes!!!

<roba> i think it looks like it will be easier to have three files and be clear about them - both for design and usage, than conflate concerns in single files..

<SimonCox> 2 possible patterns: 1. subclass to add restrictions, if they are inconsistent with the original definitions 2. merely add axioms to the sosa class, provided they are not inconsistent with the original text definitions

<Kjanowic_> so please can we vote on josh's first sentence?

<Kjanowic_> Yes, Josh is right!

<roba> clear - and consistent with the three layer pattern IMHO

<roba> can we have principle 2: axioms added to classes in a reused namespace must be completely consistent with the imported text definitions (and any imported axioms)

<Kjanowic_> so lets vote

<Zakim> kerry, you wanted to wonder what happened to me on the q and to nuance that

<Kjanowic_> Can we vote on 2+ files, 2+ namespaces, and 2+ IRIs?

<kerry> do i have the floor?

<ahaller2> yes

<kerry> I cannot follow, i am afraid, therefore I am unwilling to be pushed into a vote I do not understand

<kerry> However, can I ask whther anybody looked at the proposal with worked example that I placed on the wiuki sa few days ago?

<roba> 2+ files, 2+ IRIs -> and only need different namespaces if we need subclasses with different set membership

<kerry> there bein no answer I take that as no....

<SimonCox> kerry: can you give link please

<kerry> so can I please urge people to look at ti and coment

<ahaller2> @kerry: yes we looked, not yet at that stage in the discussion

<RaulGarciaCastro> https://www.w3.org/2015/spatial/wiki/Proposals_for_rewriting_SSN#Compromise_Proposal_6_made_by_Kerry_January_2017

<roba> kerry - i looked at it ;-)

<kerry> It is the first time a comprehensive pattern has been proposed with a worked example

<Kjanowic_> https://www.w3.org/2015/spatial/wiki/Proposals_for_rewriting_SSN#Compromise_Proposal_6_made_by_Kerry_January_2017

<RaulGarciaCastro> kerry - me too

<ahaller2> PROPOSED: SOSA and SSN new use seperate IRIs

<kerry> Ii know rob did a good job summarising an alternative pattern also on the mailing list

<SimonCox> +1

<Joshlieberman> +1

<DanhLePhuoc> +1

<Kjanowic_> +1 for PROPOSED: SOSA and SSN new use seperate IRIs

<ahaller2> +1

<roba> +1

<kerry> but frnakly i see lots of bits of issues but cno comprehensive pattern with examples

<RaulGarciaCastro> +1

<kerry> would it be possibe for an alternative (complete) pattern to be descibed and supported by examples on the wiki/

<kerry> ?

<ahaller2> @kerry, yes we do not have a resolution on how we do the integration

<kerry> o

Note from Francois, the following was initially resolved but canceled after further exchanges on IRC:
RESOLUTION: SOSA and SSN new use seperate IRIs

<ahaller2> PROPOSED: SOSA and SSN new use seperate ontology files

<SimonCox> kerry: your proposal looks basically sound to me. You can use my tables as an inventory to complete.

<Kjanowic_> +1

<kerry> -1

<kerry> -1

<kerry> when was the vote?

<kerry> -1

<Joshlieberman> +1

<DanhLePhuoc> +1

<SimonCox> +1

<ahaller2> +1

<ahaller2> kerry: why?

<kerry> I don't know what I am voting on

<roba> kerry - this was your first principle - we've just been decluttering its assumptions and conflated issues :-)

<ahaller2> on two seperate ontology files, msg: 8:55

<RaulGarciaCastro> +1

<kerry> I see a resolution that I did not vote one

<kerry> -1

<ahaller2> kerry you voted o

<kerry> pardon?

<kerry> do you mean "o"

<ahaller2> 08:54 kerry: o

<SimonCox> kerry: your proposal looks basically sound to me, particularly the section headed "ssn fragment, assuming ssn imports sosa". You can use my tables as an inventory to complete.

<kerry> i did type an "o" but it was not "0"

<kerry> and was not in response to any vote

<ahaller2> that is good enough for a resolution, everyone else +1

<ClausStadler> to me it seems reasonable having these things separated in different files, but i since i missed the discussions over the past week i can't do a educated vote

<kerry> -1

<Kjanowic_> @simoncox: as long as we do not have axioms of the type sosa:xyz [something] ssn:xyz

<kerry> I voted -1 every time


<ahaller2> @kerry: formal objection to seperate IRIs?

<kerry> formal objection to having been undersoot to vote anything other than -1

<kerry> please, it would help if i knew what vote we had?

<kerry> is it the "resolved" that was types into the irc?

<kerry> As you know, I cannot hear

<kerry> I am relying on the irc

<kerry> do i have the floor please?

<kerry> Is that yes?

<ahaller2> @kerry we are listening to Francois

<kerry> Then I want to make it very clear that every vote I have placed in this meeeting is a -1

<Joshlieberman> Yes. There were two proposals: whether to have separate ontology IRI's and whether to have separate ontology files.

<kerry> I want to make that very clear not because I am a pedantic, but because it seems it has been assumed that I have voted in favour of something (not sure what)

<Joshlieberman> We have not voted on separate namespaces.

<kerry> so... have we voted on something?

<kerry> has something been resolved?

<ahaller2> no, nothing resolved, because of your objection

<Joshlieberman> I replied to You, kerry. Can you see the IRC.

<roba> i agree we dont have a disagreement - just got the voting process too hard to follow on irc alone :-(

<roba> kerry hasnt actually intended to vote -1 for everything - just how we interpreted her irc shorthand...

<SimonCox> why not PROPOSED: SOSA elements have their own namespace. SSN elements use a different namespace, except where they re-use elements imported from other ontologies, including SOSA

<ahaller2> PROPOSED: SOSA and SSN new use seperate ontology files

<ahaller2> please vote

<SimonCox> +1

<ahaller2> +1

<roba> +1\

<Joshlieberman> +1

<Kjanowic_> Simon , please speak up

<Kjanowic_> +1

<RaulGarciaCastro> +1

<roba> +1

<DanhLePhuoc> +1

<Zakim> kerry, you wanted to answer armin

<ahaller2> @kerry we do not have a resolution on either of the two

<ahaller2> it is just to document in the minutes

<kerry> ok, then can you please ensure this above is removed from the minutes "RESOLVED: SOSA and SSN new use seperate IRIs" (Note from Francois: minutes edited afterwards to state that resolution was not adopted)

<SimonCox> This is good https://www.w3.org/2015/spatial/wiki/Proposals_for_rewriting_SSN#ssn_fragment.2C_assuming_ssn_imports_sosa

<SimonCox> Why haven't we been able to move forward?

<SimonCox> It is also largely consisten with my proposals in the last week

<kerry> to appreciate that simon has looked at the proposal

<Joshlieberman> Violent near-agreement and on that note, bye

<kerry> and also to say that Krz has expressed strong objectnions (I am not sure he read it though)

<Kjanowic_> I did

<SimonCox> (I understand K's objections though)

<kerry> Is the meeting still going?

<SimonCox> bye

<RaulGarciaCastro> Bye!

<ahaller2> @kerry, no just closed

<Kjanowic_> yes, because owl does not function that way

<kerry> ta

<kerry> bye.

<Kjanowic_> bye

<roba> i think we havent actually disagreed with the content - just couldnt get the process working smoothly :-(

<SimonCox> Make sure to capture minutes ...

<ahaller2> UNRESOLVED: SOSA and SSN new use seperate IRIs

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.148 (CVS log)
$Date: 2017/02/02 08:33:14 $