W3C home > Mailing lists > Public > public-sdw-wg@w3.org > March 2017

Re: WG discussion: shall we recommend a "samePlaceAs" property?

From: Stephane Fellah <stephanef@imagemattersllc.com>
Date: Wed, 15 Mar 2017 19:31:15 -0400
Message-ID: <CA+VBKCHD6Tdm1KRMNJAgvaiuFox1_L+Yy3DAAtvSxz07is9BbA@mail.gmail.com>
To: Jeremy Tandy <jeremy.tandy@gmail.com>
Cc: Clemens Portele <portele@interactive-instruments.de>, Rob Atkinson <rob@metalinkage.com.au>, Josh Lieberman <jlieberman@tumblingwalls.com>, "janowicz@ucsb.edu" <janowicz@ucsb.edu>, Kerry Taylor <kerry.taylor@anu.edu.au>, Bill Roberts <bill@swirrl.com>, SDW WG Public List <public-sdw-wg@w3.org>
Hi,

During OGC Testbed 10, I raised the issue related to the misuse of
owl:sameAs.

Here the section relevant (12.3.10.1) from the Engineering Report OGC-14-029
<https://portal.opengeospatial.org/files/?artifact_id=59336>

To denote that a place in a gazetteer is the ‘same’ as another one in
another gazetteer, the intuitive way is to use the *owl:sameAs* relation.
However owl:sameAs has been misused in many existing linked data due to
misunderstanding of the rules of inference defined in OWL. The following
paper discusses some of the issues with the misuse of owl:sameAs:
http://www.w3.org/2009/12/rdf-ws/papers/ws21.A

A separate property was proposed *gaz:sameLocationAs* instead. This
property is transitive and symmetric, so it will infer the mapping on other
instances.


Regards


Stephane



On Wed, Mar 15, 2017 at 2:46 PM, Jeremy Tandy <jeremy.tandy@gmail.com>
wrote:

> Yes. It's not place / location domain-specific... but the OSi example
> shows it being used in the way I was thinking for samePlaceAs.
>
> Jeremy
>
> On Wed, 15 Mar 2017 at 18:44, Clemens Portele <portele@interactive-
> instruments.de> wrote:
>
>> Jeremy,
>>
>> doesn’t "similar to" has a different meaning than "same place/location
>> as"?
>>
>> Clemens
>>
>> On 15 Mar 2017, at 18:58, Jeremy Tandy <jeremy.tandy@gmail.com> wrote:
>>
>> Hi. As agreed during the plenary call on 8-Mar, I have updated BP14 to
>> include a proposal for "samePlaceAs".
>>
>> However, having just taken a look at an example from data.geohive.ie (the
>> "Irish example" from [1]), I see use of an alternative to 'samePlaceAs':
>>
>> <http://open.vocab.org/terms/similarTo> : "Having two things that are
>> not the owl:sameAs but are similar to a certain extent. It is thought of
>> being used where owl:sameAs is too strong but rdfs:seeAlso is too loose."
>>
>> In the snippet below you can see the relationship stated to a dbpedia
>> resource:
>>
>> <http://data.geohive.ie/resource/county/2AE19629144F13A3E055000000000001>
>>       rdf:type <http://ontologies.geohive.ie/osi#County> , geo:Feature ;
>>       rdfs:label "DUBLIN"@en , "DUBLIN" , "Baile Átha Cliath"@ga ;
>>       *ov:similarTo* <http://dbpedia.org/resource/County_Dublin> ;
>>       ... ;
>>       .
>>
>> What do you think?
>>
>> (side-bar discussions already give +1 votes from Linda and Andrea)
>>
>> Jeremy
>>
>> On Tue, 28 Feb 2017 at 21:58 Rob Atkinson <rob@metalinkage.com.au> wrote:
>>
>> I think we can only point to ad-hoc, and sometimes downright bad
>> practices (owl;sameAs pointing to google maps interface.... )
>> Need to add this to the "open issues" list IMHO
>>
>> Rob
>>
>> On Wed, 1 Mar 2017 at 06:04 Joshua Lieberman <
>> jlieberman@tumblingwalls.com> wrote:
>>
>> Agreed. There is certainly interest in defining qualitative spatial
>> relationships that can be asserted and inferred even if geometrically they
>> are  imprecise or complex to calculate. However, “Place” is not just a
>> position or even a geometry, but a type of feature. samePlaceAs asserts a
>> much more detailed relationship than “collocated” or
>> “notSpatiallyDisjoint”, which may be closer to what the proposers were
>> considering.
>>
>> —Josh
>>
>>
>> On Feb 28, 2017, at 1:53 PM, Krzysztof Janowicz <janowicz@ucsb.edu>
>> wrote:
>>
>> Hi,
>>
>>
>> Generally speaking I don't think that a predicate as samePlaceAs would be
>> very useful. As far as I recall, Todd Pehle tried to introduce such
>> predicate a few years ago and it was not really used.
>>
>> First, we would also need samePersonAs, sameEventAs, and so forth, and
>> secondly, the meaning of samePlaceAs remains unclear. The issue is not only
>> that owl:sameAs is more formal in a mathematical sense (which, as stated in
>> this thread, is not always desired), it also related to URIs to each other
>> by stating that both of them point to the same feature (e.g., the same
>> place in the physical world).  What would samePlaceAs do? If it would
>> relate two places (not URIs), what does it mean for two places to be the
>> same or even similar?
>>
>> Cheers,
>> Jano
>>
>>
>>
>> On 02/28/2017 02:38 AM, Kerry Taylor wrote:
>>
>> +1
>>
>> *From:* Jeremy Tandy [mailto:jeremy.tandy@gmail.com
>> <jeremy.tandy@gmail.com>]
>> *Sent:* Tuesday, 28 February 2017 2:11 AM
>> *To:* Bill Roberts <bill@swirrl.com> <bill@swirrl.com>; SDW WG Public
>> List <public-sdw-wg@w3.org> <public-sdw-wg@w3.org>
>> *Subject:* Re: WG discussion: shall we recommend a "samePlaceAs"
>> property?
>>
>> Thanks Bill.
>>
>> > Probably a better option would be to propose it to danbri for addition
>> to schema.org as a property for things of type schema:Place ?
>>
>> You're right that that sounds like a better home.
>>
>> @danbri: what do you think? (& can you remind us how we might propose
>> this for schema.org's consideration)
>>
>> Thanks. Jeremy
>>
>>
>> On Mon, 27 Feb 2017 at 13:43, Bill Roberts <bill@swirrl.com> wrote:
>>
>> I support creating a samePlaceAs relation. As well as an IANA link
>> relation, can we have a URI for it to allow use in RDF?
>>
>> Possibly related, I see in BP10 that we refer to ongoing work to update
>> GeoSPARQL - what's the status of that? Would this property/relation make
>> sense as part of the new GeoSPARQL? Maybe the deliberate vagueness of
>> 'samePlaceAs' might not fit well with the otherwise generally precise
>> geosparql relationships.
>>
>> Probably a better option would be to propose it to danbri for addition to
>>  schema.org as a property for things of type schema:Place ?
>>
>>
>>
>> On 27 February 2017 at 11:44, Jeremy Tandy <jeremy.tandy@gmail.com>
>> wrote:
>>
>> Hi - for this sprint in development of the Best Practice document, we're
>> updating BPs about "linking" and "vocabularies" ...
>>
>> On multiple previous occasions (most recently the London F2F) we've
>> mentioned that we should propose a "samePlaceAs" property. In essence, I
>> think we see this as a subjective statement (that a human might make)
>> rather than a mathematical / topological statement, matching on the spatial
>> characteristics only.
>>
>> This addresses the concerns about the VERY restrictive owl:sameAs. At
>> TPAC2016, @clemens said that a "relaxed relationship is better [for
>> cross-referencing identifiers] (e.g. samePlaceAs) … but if you _can_ state
>> owl:sameAs then you should do so … " [from my notes]
>>
>> We said at TPAC2015 "samePlaceAs would be a 'social relationship' - based
>> on people's perception".
>>
>> The domain and range should both be "spatial things" (which definition of
>> spatial thing do we refer to - the new one coming from @josh's work or
>> w3cgeo:SpatialThing?
>>
>> We're looking to resolve this question BEFORE the Delft F2F.
>>
>> WG members: what do you think?
>>
>> Many thanks, Jeremy
>>
>> further notes below:
>>
>> ---
>>
>> My notes from the most recent discussion during London F2F are here:
>>
>>    - "samePlaceAs"
>>    - it would be an IANA link relation identifier
>>
>>
>>    - equivalence at a geographical level - without a formal definition
>>       of that equivalence
>>       - geography related
>>       - don't express as a sub-property of, for example, "so:matches" [
>>       <https://files.ifi.uzh.ch/ddis/iswc_archive/iswc/pps/web/iswc2010.semanticweb.org/pdf/261.pdf>
>>       https://files.ifi.uzh.ch/ddis/iswc_archive/iswc/pps/
>>       web/iswc2010.semanticweb.org/pdf/261.pdf
>>       <https://files.ifi.uzh.ch/ddis/iswc_archive/iswc/pps/web/iswc2010.semanticweb.org/pdf/261.pdf>]
>>       … we're only indicating that the _spatial_ properties match ... and
>>       property hierarchies just get complicated to 95% of humans
>>       - not a mathematical statement (like the topological relationships)
>>       - avoid any "mereological" confusion [ :) ]
>>       - nearby? (& other fuzzy relationships) ... same-place-as is _so_
>>       common that we'll deal with it as a special case and _not_ cover these
>>       other spatial relationships for now
>>       - which ontology? IANA Link Relations
>>
>>
>>    - ... not used today- so not a _best_ practice
>>    - ... assert as a [recommended] approach to resolve problems we see
>>    in evidence today- especially regarding incorrect use of owl:sameAs
>>
>> ---
>>
>> And back on the BP call on 9-Nov we said:
>>
>> *jtandy:* Another aspect to discuss is the reuse of identifiers ("to
>> keep the global graph intact").
>>
>> ... But to be able to add additional information and make it retrievable
>> it requires a new identifier with a sameAs-like link to the "known
>> identifier"
>>
>> ... "samePlaceAs"?
>>
>> *eparsons:* samePlaceAs sounds restrictive
>>
>> *jtandy:* agrees, we want to avoid the strong nature of sameAs
>>
>> *ByronCinNZ:* likes the idea, very geographic statement. In which
>> ontology would this reside?
>>
>> *ClemensPortele:* I think we said it would be an IANA link relation
>> identifier
>>
>> *jtandy:* As it does not exist yet, we cannot claim it is a "best
>> practice"
>>
>> *eparsons:* I think this problem will be hard to avoid, but it could be
>> described as a way to address the issue
>>
>> *ChrisLittle:* worried about "samePlaceAs". How does it fit with the
>> algebra of polygons?
>>
>> *jtandy:* we don't want to be too specific
>>
>> ... ... at TPAC we had a discussion about the well-defined topological
>> relationships
>>
>> *eparsons:* to get something done quickly we should try to keep it simple
>>
>> ... ... relationships could be tackled later
>>
>> *jtandy:* so we agree that samePlaceAs is not intended as a mathematical
>> statement
>>
>> <*ClausStadler_*> https://files.ifi.uzh.ch/ddis/iswc_archive/
>> iswc/pps/web/iswc2010.semanticweb.org/pdf/261.pdf
>>
>> <*ClausStadler_*> "so:matches Two URIs refer to possibly distinct things
>> that share all the prop- erties needed to substitute for each other in some
>> graphs. Th is property is symmetric but not necessarily reflexive.
>> so:matches is a super-property of so:identical ."
>>
>> *ByronCinNZ:* agrees, and this is probably the most important of the
>> topological relationships
>>
>> *ClausStadler_:* Explains the paper and "so:matches" reference (see
>> above)
>>
>> *jtandy:* yes, there is overlap. we want to focus on the spatial match.
>>
>> *ClausStadler_:* could be a sub-property
>>
>> *jtandy:* worried on nesting, maybe it makes it overcomplicated
>>
>> I agree with the concern
>>
>> <*eparsons*> +1
>>
>> *eparsons:* worried about complication, too
>>
>> <*AndreaPerego*> +1
>>
>> *ByronCinNZ:* should be a top-level relationship
>>
>>
>>
>>
>>
>> --
>> Krzysztof Janowicz
>>
>> Geography Department, University of California, Santa Barbara
>> 4830 Ellison Hall, Santa Barbara, CA 93106-4060
>>
>> Email: jano@geog.ucsb.edu
>> Webpage: http://geog.ucsb.edu/~jano/
>> Semantic Web Journal: http://www.semantic-web-journal.net
>>
>>
>>
>>


-- 
Stephane Fellah
Chief  Knowledge Scientist
Image Matters LLC
Office: +(703) 669 5510
Cell: 703 431 9420
Received on Wednesday, 15 March 2017 23:31:49 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 15 March 2017 23:31:50 UTC