Re: On LDOM

On 1/28/15 2:49 PM, Holger Knublauch wrote:
> Hi Karen,
>
> it is obvious from reading the various emails that I need to do a better
> job explaining what LDOM really is. It is all too easy to draw false
> conclusions. I can write down the same train of thoughts over and over
> again, yet they don't seem to "click". Maybe it is because emails and
> text documents are just not the right medium. I


I think that some examples, with human-readable documentation if they 
are in code (for those of us who do not easily read code), would help. 
Eric's examples could be a good model. It would be useful to see:

the ontology(s) describing the classes and properties
the instance data
the LDOM code
the result of the LDOM validation

'd be more than happy to
> talk to any WG member on the phone if that helps (just contact me
> off-list). In the absence of better communication channels, let me try
> to respond to your specific concerns here.
>
> On 1/29/2015 5:07, Karen Coyle wrote:
>> I see. I was reading it as a model for managing and validating RDF or
>> OWL information in private/closed data stores. In fact, it seems to
>> eschew any contact with the web,
>
> Why do you believe that a language called "Linked Data Object Model"
> eschews any contact with the Web. As written in the first sentence of
> the Primer
>
> "LDOM (Linked Data Object Model) is an RDF-based modeling language that
> is compatible with Linked Data principles and leverages some
> object-oriented concepts to the Web."

I'm not at all clear on what it means that it "leverages some 
object-oriented concepts to the Web."

>
> LDOM works very well on both closed and open networks. Class, shape and
> property definitions as well as templates and functions can be looked up
> from the web using the normal RDF resolution mechanisms. It is
> extensible - people can reuse each other's data models across web sites
> and applications. It has exactly the same characteristics that RDFS and
> OWL have, only that it uses the Closed World assumption, and this is
> exactly what the WG was chartered to deliver.

As long as none of that CWA trickles out to the OW that I wish to 
operate in. In other words, I'll want a CWA in my private data store, 
but on the web I need to operate completely within the OW rules. Which 
is why I continue to ask whether one set of instance data will be 
appropriate to both. I don't want to design a class structure that is 
aimed at my validation rules but that will have a different OW meaning. 
An easy example is disjointness -- I may want some properties or classes 
to be disjoint for validation, but not carry that into the OW because it 
could interfere where I wish to mix my data with that of others. I want 
to keep validation and RDF semantics separate.

>
>> and treats the semantic aspects of RDF/OWL as (and I quote Holger)
>> "pie in the sky."
>
> You are quoting me out of context. I never said this about the semantic
> aspects of RDF/OWL. I said this about the idea of mixing together all
> triples in the world in an uncontrolled setting. That vision is of
> course completely unrealistic, because anyone could insert a single
> triple that basically invalidates all other statements. This isn't a
> stable architecture that anyone in practice would use, unless perhaps
> for browsing applications where a human is always in the loop.

And yet, essentially, for us the open web is where our users are. We 
have to go there. Therefore we are very oriented toward that solution.

>
>> I'm still trying to understand if RDF data intended for the web will
>> fit into its processing model, and again if local data that can be
>> managed by it could also have a web presence without great modification.
>
> Yes. LDOM can obviously be used in closed scenarios. In open scenarios
> it can, for example, be used to publish reusable ontologies that other
> applications and ontologies can reuse. Publishing an LDOM file on the
> Web then means that anyone who wants to reuse the terms from that file
> *should* adhere to the attached semantics. This is exactly like RDFS/OWL
> also do it.

I guess that depends on what you mean by the "attached semantics". 
Classes are optional, and any "thing" can be an instance of multiple 
classes - meaning that others are free to associate classes to my 
subjects if it serves their need. In DCMI, an application profile (which 
is a precursor to RDF validation, conceptually) is a description of what 
meaning/shape I confer on my data. What people do with the data is up to 
them. There are no constraints on reuse.


  If someone just wants to publish a vocabulary without such
> constraints, they can already do that by creating RDFS classes with
> comments, and put the constraints into comments and documentation (as
> done, for example, for SKOS, RDF Data Cube and PROV). LDOM just makes
> those implicit assumptions explicit, for applications that want to
> ensure better integrity of their data. It is therefore a completely
> incremental improvement over existing practices and standards. People
> who prefer to leave their data model open don't need to use LDOM at all.

It's not one or the other, but both. Within my own bit of world where I 
have control, I do want to have a way to validate data as it is created, 
and as I ingest it from other sources. I also want to be able to 
describe to others what I think my data looks like. Among "partners" we 
will agree on a set of rules. (This is sounding very close to the OSLC 
work flow.) But I also want my data on the open web, where validation 
will not take place. And this is why my question is whether that means I 
need two separate instance data definitions, or if I can make do with 
one. If I must manage validation using classes that identify the 
"validatable units" then I probably will need at least two data models 
-- one internal to my closed system, and another for the open world. And 
that's why I ask that question about open/closed worlds - knowing, 
however, that there may not be a yes/no answer, but an "it depends."

kc



>
>> I personally was hoping for a solution that is more compatible with
>> RDF data on the web.
>
> What is missing, where is LDOM not compatible with RDF data on the web?
>
> Thanks,
> Holger
>

-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
m: 1-510-435-8234
skype: kcoylenet/+1-510-984-3600

Received on Thursday, 29 January 2015 01:34:15 UTC