Re: Linked Data Templates paper for XML London 2016

I like the use of PATCH. I can imagine some LDT's allowing levels of
trust where one of the levels allows at least an "owl:sameAs" patch to
connect the dots between remote data sets.

On Thu, Mar 17, 2016 at 1:02 PM, Martynas Jusevičius
<martynas@graphity.org> wrote:
> LDT is a client-server architecture, where processor is the server,
> and client uses hypermedia. However processor and client can built
> into one node, which are like distributed Linked Data agents.
>
> If I'm getting your workflow, I think this be accomplished with 2
> requests. I've omitted the prefixes.
>
> First POST to your own server:
>
> POST http://patrick.org
>
> _:sameAs a foaf:Document ;
>   owl:sameAs <http://martynas.org/some-resource> .
>
> which gives you a skolemized URI:
>
> HTTP/1.1 200 OK
>
> <http://patrick.org/7a6eb95f-5fe1-4e3f-baf5-3314ad366a7e> a foaf:Document ;
>   owl:sameAs <http://martynas.org/some-resource> .
>
> Then you can do a PUT to my http://martynas.org/some-resource with an
> updated payload (with the extra triple) - or possibly only PATCH the
> change, like this:
>
> PATCH http://martynas.org/some-resource
>
> <http://martynas.org/some-resource> owl:sameAs
> <http://patrick.org/7a6eb95f-5fe1-4e3f-baf5-3314ad366a7e> .
>
> Wouldn't that be more RESTful?
>
> On Thu, Mar 17, 2016 at 8:24 PM, Patrick Logan <patrickdlogan@gmail.com> wrote:
>> I am thinking of something like...
>>
>> 1. You are running an LDT and have a resource at
>> http://martynas.org/some-resource
>>
>> 2. I would like to enrich this data with my own data, using my own LDT.
>>
>> 3. I need my own IRI on my server to do so. I might do something like
>> the following:
>>     POST http://patrick.org/mintOwlSameAs?url=http://martynas.org/some-resource
>>
>> 4. This would return some local IRI I could use, like
>> http://patrick.org/some-resource
>>
>> The data set on my LDT would have a triple like:
>>
>> <http://patrick.org/some-resource> owl:sameAs
>> <http://martynas.org/some-resource>
>>
>> 5. A next step could be "federation" where my server might request
>> that your server also add the owl:sameAs and provide my IRI.
>>
>> On Thu, Mar 17, 2016 at 12:02 PM, Martynas Jusevičius
>> <martynas@graphity.org> wrote:
>>> Patrick,
>>>
>>> great to have feedback :)
>>>
>>> Could it be that with "minting" you have "skolemization" in mind?
>>> https://github.com/Graphity/graphity-processor/wiki/Data-input#blank-node-skolemization
>>>
>>> On Thu, Mar 17, 2016 at 7:28 PM, Patrick Logan <patrickdlogan@gmail.com> wrote:
>>>> Thinking about this just a little bit more... I wonder whether an
>>>> additional operation could be useful...
>>>>
>>>> ...a POST to "mint" an IRI on the current LDT host that is an OWL
>>>> "same as" for a given IRI. The original IRI may or may not be hosted
>>>> on some other LDT system. The new IRI could then be used for local LDT
>>>> operations about that resource.
>>>>
>>>>
>>>> On Thu, Mar 17, 2016 at 10:53 AM, Patrick Logan <patrickdlogan@gmail.com> wrote:
>>>>> I like the conceptual regularity of IRI's, pattern matching, and
>>>>> templates and the HTTP verbs.
>>>>>
>>>>> My main question would be about the range of applications that can be
>>>>> built with respect to using standard vocabularies. i.e. my
>>>>> understanding is I am limited in applying the HTTP verbs only to
>>>>> resources that share the IRI prefix with the linked data host. The
>>>>> templates can refer to all the standard and common vocabularies, but I
>>>>> cannot PUT or POST anything about a FOAF person for example, unless
>>>>> that person has an IRI on that LDT host?
>>>>>
>>>>> As I said, it makes neat conceptual sense. Could it limit the range of
>>>>> application expressiveness? Could there be "patterns of usage" that
>>>>> would allow a kind of "federation of LDT hosts" that would support
>>>>> federating your server and my server to talk about the same resources
>>>>> (that would have one IRI for the resource as it is on your server and
>>>>> another IRI for the same resource as it is on my server?
>>>>>
>>>>> This is definitely a good step forward for the LD platform definition.
>>>>>
>>>>>
>>>>> On Wed, Mar 16, 2016 at 10:24 AM, Martynas Jusevičius
>>>>> <martynas@graphity.org> wrote:
>>>>>> Hey all,
>>>>>>
>>>>>> we have submitted an extended abstract for the XML London 2016 conference:
>>>>>> http://xmllondon.com
>>>>>>
>>>>>> We will be notified on the 7th of April whether it gets accepted. If
>>>>>> it does, we will need to write the final paper. Fingers crossed :)
>>>>>>
>>>>>> The abstract goes like this:
>>>>>>
>>>>>> Linked Data Templates define the syntax and the semantics of a Linked
>>>>>> Data processor which publishes and consumes RDF data over HTTP. The
>>>>>> processor responds to Linked Data requests by interpreting a sitemap
>>>>>> ontology as instructions to indicate how the request metadata maps to
>>>>>> an operation on SPARQL service, and how to generate response body. The
>>>>>> LDT vocabulary also provides capabilities to define hypermedia
>>>>>> controls, container resources with paginated access, resource
>>>>>> constructor templates, validation constraints and skolemization
>>>>>> templates.
>>>>>>
>>>>>> You can find the whole document here:
>>>>>> https://docs.google.com/document/d/1uUIkSKQly-td7F9QjXS7QE-lzWL3ytGxDd5dzmvuO1c/edit?usp=sharing
>>>>>>
>>>>>> Feedback is very welcome. You can comment in the Google doc as well.
>>>>>>
>>>>>> We are currently working on the draft on the specification, expecting
>>>>>> to make it public next month.
>>>>>>
>>>>>>
>>>>>> Martynas
>>>>>> graphityhq.com
>>>>>>
>>>>
>>

Received on Thursday, 17 March 2016 20:27:25 UTC