Replies: 29 comments
-
|
Hi Enrico (@ecow)! We are not using the URI to mean two different things. We dereference URIs using content negotiation mechanisms to provide different views on URIs for different agents. |
Beta Was this translation helpful? Give feedback.
-
|
I do not understand. How you differentiate the web Page from the FIBO concept? In other words if you say: https://spec.edmcouncil.org/fibo/ontology/FND/Accounting/CurrencyAmount/AmountOfMoney dcterm:created "20200701" . Are you referring the date of creation of the FIBO concept or the date of the creation of the Page describing it? To use the same uri for two different meanings (a web Page is a foaf:Document) breaks web standards. IMHO, this has nothing to do with content negotiation. |
Beta Was this translation helpful? Give feedback.
-
|
Everything in the FIBO code applies to the ontological resources (concepts, properties, etc.). It has nothing to do with the FIBO Viewer and the web pages. |
Beta Was this translation helpful? Give feedback.
-
|
I'm a bit confused by this discussion - it seem to have started with the observation "[EDMC] are using the same URI for the web page ... and for the abstract concept ..." I'm not sure what this observation is based on; if somewhere we have a triple that says something like: https://spec.edmcouncil.org/fibo/ontology/FND/Accounting/CurrencyAmount/AmountOfMoney a ex:WebPage . then it is indeed an error. I haven't been able to find any such triple. If it has to do with the content negotiation (despite Enrico's comment that it doesn't), that is, the fact that if you resolve https://spec.edmcouncil.org/fibo/ontology/FND/Accounting/CurrencyAmount/AmountOfMoney with a web browser, you get a web page, then this seems indeed to be mistaking content negotiation for a statement of type. It is pretty common practice (and many FIBO users for years scolded us for not doing this) to have RDF resources resolve to web pages about themselves when resolved on behalf of a web browser. For example, the SKOS URL http://www.w3.org/2004/02/skos/core# resolves to a web page about the standard when resolved using a web browser, and resolves to an RDF/XML file when resolved using an appropriate media type (e.g., application/rdf+xml). Similar comments apply to Dublin Core terms (http://purl.org/dc/terms/) and XML Schema ( http://www.w3.org/2001/XMLSchema#) FIBO resources resolve to RDF fragments in appropriate serializations when fetched with corresponding media types as outlined in https://www.w3.org/2008/01/rdf-media-types; For example, curl -H "Accept: application/rdf+xml" https://spec.edmcouncil.org/fibo/ontology/FND/Accounting/CurrencyAmount/AmountOfMoney returns triples in RDF/XML format We have done our best to conform to W3C practice, by imitating what they themselves have done for similar efforts (e.g., SKOS, XSD) and what other have done in similar contexts (e.g., Dublin core). We have also responded to complaints by others that by failing to resolve FIBO resources to human readable pages when resolved on behalf of a web browser, we were out of compliance with best web practices. If there is some way in which we have fallen short, please help us to understand what we should do differently. But I don't see how we have made a statement that AmountOfMoney is a web page any more than the W3C has made a statement that http://www.w3.org/2004/02/skos/core# is a web page. They explicitly state that http://www.w3.org/2004/02/skos/core# a owl:Ontology . just as we explicitly state fibo-fnd-acc-cur:AmountOfMoney a owl:Class . |
Beta Was this translation helpful? Give feedback.
-
|
I understand, But anyway you publish web pages that use the same URI. To be consistent with web standard you must use a different URI to represent the FIBO concept and the related FIBO Viewer web page. for instance:
Another viable solution could be to use hash notation, i.e.: - or service notation:
As an example, the dpedia.org choice was to use:
Dereferencing without generating collision is one of the main requirement in the Web and Linked Data standards. For more info see https://www.w3.org/TR/webarch/#URI-collision I think this question should be moved to an issue. Thank you for your support |
Beta Was this translation helpful? Give feedback.
-
|
Ps @dallemang look well:
and it is dereferenced to http://www.w3.org/2004/02/skos/core is an ontology (i.we. a non informative web resource), It seams a confusion between "dereferencing" and "content negotiation", they are different things that address different problems in web architecture and in http protocol specification. http://www.w3.org/2004/02/skos/core# is a name space without any formal meaning (for what i understand) |
Beta Was this translation helpful? Give feedback.
-
|
@ecow, thank you for your comments. Of course, you can use the service notation (URLs with a query string) to refer to the FIBO web pages (e.g.: I asked @przemekgradzki to analyze the situation. We will soon find out what his opinion is. |
Beta Was this translation helpful? Give feedback.
-
|
@ecow , @dallemang , @trypuz: what do you think about this solution:
I think, that correct URI for "RDF document" is: |
Beta Was this translation helpful? Give feedback.
-
|
I like your first and third URI. One might consider, instead of https://spec.edmcouncil.org/fibo/ontology/viewer/FND/Accounting/CurrencyAmount/AmountOfMoney to use https://spec.edmcouncil.org/fibo/viewer/FND/Accounting/CurrencyAmount/AmountOfMoney since this entity is not really in the ontology at all; it is a human-viewable entity. As analogy, the URIs is the Vocabulary have names like .../fibo/vocabulary/... and in the data dictionary .../fibo/datadictionary/... , that is, the product name comes right after fibo. Since the viewer is also a product, wouldn't it make sense to put its name at that location? |
Beta Was this translation helpful? Give feedback.
-
|
@dallemang good point |
Beta Was this translation helpful? Give feedback.
-
|
Yea, all proposed solution works well. My suggestion is to use the Recipe 5 in the W3C document Best Practice Recipes for Publishing RDF Vocabularies so, the than The only unavoidable drawback is that someone could be tempted to cut&and paste the link in the the browser URL bar and try to use it instead of the concept URI. But this risk is already well mitigated by the presence of the [COPY] button in the viewer. This is a well known topic :-) |
Beta Was this translation helpful? Give feedback.
-
|
I have always thought it very odd that any “pure” identifier would include a technical protocol (http:/https:) rather than something to indicate that it is just identity without expectation of access, like “id:”.
Where an access protocol is specified, it seems reasonable for that identifier to indicate where information about that identity can be found, at least information published by the authority minting the URI. It does not need to mean that the URI is the identity of that referenced resource and the thing, just that the identity of the thing can be dereferenced to information about the thing. Without any link between identity and information we would be lost. Pure identity is nice in theory but impractical in any open system.
…-Cory
From: Dean Allemang <[email protected]>
Sent: Tuesday, July 7, 2020 10:23 AM
To: edmcouncil/fibo <[email protected]>
Cc: Subscribed <[email protected]>
Subject: Re: [edmcouncil/fibo] Concept URI vs dereferenced web page URI (#1072)
I like your first and third URI.
One might consider, instead of
https://spec.edmcouncil.org/fibo/ontology/viewer/FND/Accounting/CurrencyAmount/AmountOfMoney
to use
https://spec.edmcouncil.org/fibo/viewer/FND/Accounting/CurrencyAmount/AmountOfMoney
since this entity is not really in the ontology at all; it is a human-viewable entity. As analogy, the URIs is the Vocabulary have names like .../fibo/vocabulary/... and in the data dictionary .../fibo/datadictionary/... , that is, the product name comes right after fibo. Since the viewer is also a product, wouldn't it make sense to put its name at that location?
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub<#1072 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABIJNJDDKU6PCFJYWLU5LC3R2MVUJANCNFSM4OOUNYSQ>.
|
Beta Was this translation helpful? Give feedback.
-
|
I agree with @trypuz 's initial reaction, @dallemang 's line of questioning, and @CoryCasanave's last remark - though I'm not so bothered at the inclusion of protocol in the URI. .. IMO with FIBO the situation is not the equivalent of Milan: we don't have a separately curated web page about the concept - the page is just a query of all the things we know about the concept. FIBO is all concepts, it's not about managing web pages, as Dean said. |
Beta Was this translation helpful? Give feedback.
-
|
The question is: does the current setting (including FIBO Viewer config, Nginx config, and FIBO) break any W3C/RFC/SW standard? |
Beta Was this translation helpful? Give feedback.
-
|
@trypuz 3. URIs for Real-World Objects: The most important is: ...one URI can't stand for both a Web document and a real-world object... Now URI=
I think it's acceptable to have one URI for web page (2.) and RDF document (3.) - they are distinguished |
Beta Was this translation helpful? Give feedback.
-
|
@przemekgradzki Suppose you have a URL that identifies a John Doe, a real person. No curl request will return John Doe himself, but only a document or file that somehow names or identifies him. But this cannot be raised as an argument that the URL in question identifies both John Doe and some file/document about him. All URLs direct to web resources and not to real world objects (well at least not until we end up in the Matrix), even if some of them identify real world objects. The fact that a given curl request returns an html document, is not relevant for the question what this request's url identifies. In the case of FIBO the situation is more convoluted because FIBO consists of concepts and these are not like persons. Still, I think my point above stands also for FIBO as these concepts are probably not just parts of rdf/xml documents. |
Beta Was this translation helpful? Give feedback.
-
|
@mereolog
...one URI can't stand for both a Web document and a real-world object...
And stands for does not mean is |
Beta Was this translation helpful? Give feedback.
-
|
I am confused by point (3); this URI has "/viewer/" in it, and therefore stands for the web document (which, as @rivettp pointed out, does not even exist). The URI for the concept of AmountOfMoney is https://spec.edmcouncil.org/fibo/ontology/FND/Accounting/CurrencyAmount/AmountOfMoney (notice that /viewer/ is missing) |
Beta Was this translation helpful? Give feedback.
-
|
@dallemang - typo in URI 3.; fixed |
Beta Was this translation helpful? Give feedback.
-
|
@przemekgradzki |
Beta Was this translation helpful? Give feedback.
-
So URI identifies resource and using that URI you can get "a representation of the target resource" = web page.
Cool URIs for the Semantic Web is a practical and simple method (RFC7231, section 6.4.4. 303 See Other redirect) to link "pure identifiers" of FIBO concepts with URIs, which identifies web pages. But it's just proposal |
Beta Was this translation helpful? Give feedback.
-
|
We created this IRI scheme for FIBO years ago based on a number of ideas, one of which was the "follow your nose principle" where a user (usually a data modeler, SME or ontologist) would have to be able to click on an ontology IRI or a property IRI and, depending on the client-app from which he/she makes the click and the preferred MIME type of that client-app, would end up with the right content. If you dereference from Protege or Topbraid Composer you'd end up fetching RDF. If you dereference from a browser you'd end up HTML. Multiple "serialization formats" of the same information. And yes that is a slight deviation from the linked data standard, I agree. Done on purpose at that time (by me). One alternative at that time was to redirect in the frontend NGINX server, if you come in with a GET and an Accept header of text/html (like a browser would do) then redirect to the same IRI + |
Beta Was this translation helpful? Give feedback.
-
|
@przemekgradzki about your proposal:
Here's the original documentation of the "IRI scheme": https://github.com/edmcouncil/fibo/blob/master/etc/process/iri-scheme.md The word |
Beta Was this translation helpful? Give feedback.
-
|
That would then also allow anyone else to come up with other types of "viewers" (maybe an editor even as some point? Or a collaborative ontology environment or some product of some partner?) |
Beta Was this translation helpful? Give feedback.
-
|
Let's call the new product
This is also in line with what @dallemang suggested in this comment If you use ontology/property IRI in your web browser you will be redirected to the web page - you will see the web page URL in the address bar Is it acceptable? |
Beta Was this translation helpful? Give feedback.
-
|
No I don't like the "ontoviewer" in the URI.
|
Beta Was this translation helpful? Give feedback.
-
|
Hi folks! Please use http://linkeddata.uriburner.com:8000/vapour to validate that resolution is done correctly, as listed at http://vocab.getty.edu/doc/#Semantic_Resolution. @ecow thanks for bringing this up! |
Beta Was this translation helpful? Give feedback.
-
|
Looks OK. Do we close this issue? |
Beta Was this translation helpful? Give feedback.
-
|
Sorry, I'm not so sure that it looks good...I read:"Conclusions on the type
of the resources
- <
https://spec.edmcouncil.org/fibo/ontology/BE/LegalEntities/LegalPersons/LegalEntity>
identifies a Web document
"
This states that the LegalEntity URI is a Web document. I do not think it
is what FIBO wants to assert.
Il mer 18 nov 2020, 11:59 Robert Trypuz <[email protected]> ha
scritto:
…
http://linkeddata.uriburner.com:8000/vapour?uri=https%3A%2F%2Fspec.edmcouncil.org%2Ffibo%2Fontology%2FBE%2FLegalEntities%2FLegalPersons%2FLegalEntity&defaultResponse=dontmind&userAgent=http%3A%2F%2Flinkeddata.uriburner.com%3A8000%2Fvapour%23this
Looks OK. Do we close this issue?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#1072 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABMFIEMHHYLC4IJQGSYFZ4LSQOSJXANCNFSM4OOUNYSQ>
.
|
Beta Was this translation helpful? Give feedback.
-
Hi,
I'm developing semantic web application using some FIBO concepts and I have a question about URI dereferencing in FIBO:
It seems to me that you are using the same URI for the web page (i.e. informative resource) and for the abstract concept (i.e. non-informative resource). If this is true it could lead to a semantic collision according with the Web architecture specifications (cfg. https://www.w3.org/TR/webarch/#URI-collision and https://www.w3.org/TR/cooluris/#distinguishing). I'm wrong? How do you manage this?
e.g.:
is the dct:title a property of the https://spec.edmcouncil.org/fibo/ontology/FND/Accounting/CurrencyAmount/AmountOfMoney
URI subject referring to the web page title or of the fibo-fnd-acc-cur:AmountOfMoney conceptual owl:Class?
How can I disambiguate the two things?
Many thanks for your support
Best regards
Enrico Fagnoni
Beta Was this translation helpful? Give feedback.
All reactions