Re: SPARQL for Web Forms. was: HTML Form to semantic web proposal

> On 27 Nov 2015, at 15:44, Martynas Jusevičius <martynas@graphity.org> wrote:
> 
> A new language? Really? Like 95% SPARQL, but incompatible?

yes. It's not the same thing. It can re-use a lot of what is there, but
clearly we are not querying a graph. It may just require to be a subset of 
SPARQL. SPARQL itself is a variation of N3. And SPARQL PATCH is itself
an extension of SPARQL. This is not the first time this happens.

> 
> This way you are aiming for the anecdotal xkcd 15 incompatible
> standards situation. That is exactly what is wrong with the
> semantic/LD ecosystem.
> There is a specification that does exactly what you want: encode HTML
> form data as RDF. It is called RDF/POST and is well specified.
> Essentially it's just another RDF syntax, we even have written a
> streaming parser for it.
> 
> So the technology is not the issue. The issue is the "IndieWeb folks"
> (whatever IndieWeb is?) that are unwilling to, or unable to, learn new
> technologies that are best suited for the problem at hand. We need to
> educate developers to reuse existing specifications, not create new
> ones at every turn.
> 

Don't really want to get in a discussion about all of that. I have other
things to do. This was just a thought. It may interest people, other not.

> 
> Martynas
> 
> On Fri, Nov 27, 2015 at 3:26 PM, henry.story@bblfish.net
> <henry.story@bblfish.net> wrote:
>> 
>>> On 27 Nov 2015, at 14:21, Martynas Jusevičius <martynas@graphity.org> wrote:
>>> 
>>> How would determine the datatypes of the values, or which values
>>> should be literals and which URIs? You can only do that in the WHERE
>>> clause (using BIND() for example), and it has to operate over an RDF
>>> graph. Or WHERE has to be empty as I suggested before.
>> 
>> IN SPARQL the query has to operate over a GRAPH, but this is meant to
>> be a new language insipired and re-using as much of the work from SPARQL
>> as possible. Ideally one could take a few subsections of sparql and
>> produce it.
>> 
>> You are right that one needs to find a way to deal with datatypes.
>> I point out in a number of places in
>> https://github.com/w3c-social/webmention/issues/11
>> 
>> It would have been easy if one had been able to just specify the
>> datatype in the CONSTRUCT.
>> 
>> 
>> @prefix foaf: <http://xmlns.com/foaf/0.1/>.
>> @prefix contact: <http://www.w3.org/2000/10/swap/pim/contact#> .
>> 
>> CONSTRUCT {
>> []  foaf:name ?name;
>>    foaf:age ?age^^xsd:integer;
>>    contact:home
>>            [ a contact:ContactLocation;
>>                contact:address [ contact:city ?city;
>>                                  contact:country ?country;
>>                                  contact:postalCode ?zip;
>>                                  contact:street ?street
>>                                ] .
>> } WITH ?name ?age ?city ?country ?zip ?street
>> 
>> 
>> So yes, this is just an initial proposal, it is not fully fleshed out,
>> and one needs to see if the IndieWeb Camp folks actually would even want
>> to play along. But as I spent a few hours on this, perhaps others may
>> have other use cases for which they  would find such a tool useful.
>> 
>> 
>> 
>>> 
>>> On Fri, Nov 27, 2015 at 3:17 PM, henry.story@bblfish.net
>>> <henry.story@bblfish.net> wrote:
>>>> 
>>>> On 27 Nov 2015, at 14:13, Martynas Jusevičius <martynas@graphity.org> wrote:
>>>> 
>>>> Well then you need a canonical mapping from www-encoded data to RDF before
>>>> you query. CONSTRUCT could work on that, but it will not help you define it.
>>>> 
>>>> The only existing one I know is RDF/POST.
>>>> 
>>>> 
>>>> I don't agree that you need a canonical mapping from www-encoded data to
>>>> rdf.
>>>> 
>>>> You can think of the attribute values as the response-set of a missing
>>>> WHERE.
>>>> So you only actually need the CONSTRUCT part.
>>>> 
>>>> You do of course need the resource you are POSTing to, to return something
>>>> that
>>>> points to the transormation. This could be done with a Link header such as
>>>> 
>>>> 200 Ok
>>>> Link: <http://w3c.org/social/WebMention>; rel="urlencoded"
>>>> 
>>>> 
>>>> as explained in https://github.com/w3c-social/webmention/issues/11
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> On Fri, 27 Nov 2015 at 14:53, henry.story@bblfish.net
>>>> <henry.story@bblfish.net> wrote:
>>>>> 
>>>>> 
>>>>>> On 27 Nov 2015, at 11:17, Martynas Jusevičius <martynas@graphity.org>
>>>>>> wrote:
>>>>>> 
>>>>>> Where did you find the CONSTRUCT ... WITH query form though?
>>>>> 
>>>>> I just invented it, a quick first idea for people to get the gist.
>>>>> Clearly what would be needed would be also some way of having optional
>>>>> form attributes.
>>>>> 
>>>>>> 
>>>>>> It could probably be just CONSTRUCT template with an empty WHERE {} on
>>>>>> which you apply the variable bindings.
>>>>> 
>>>>> I don't think that would work. We are actually just doing a construct on
>>>>> an result set. The parameter values in the www-encoded POSTed content
>>>>> can be thought of as a result set.
>>>>> 
>>>>> 
>>>>> 
>>>>>> 
>>>>>> On Thu, Nov 26, 2015 at 7:53 PM, henry.story@bblfish.net
>>>>>> <henry.story@bblfish.net> wrote:
>>>>>>> That requires people getting people making such forms to learn RDF,
>>>>>>> which if
>>>>>>> you
>>>>>>> read the types of answers to issue 11 linked below, or if you interact
>>>>>>> with
>>>>>>> IndieWeb folks,
>>>>>>> you'll understand is unlikely.  What is less unlikely is that the
>>>>>>> server
>>>>>>> provide a link
>>>>>>> to an interpretation. The interpretation can then be made machine
>>>>>>> readable
>>>>>>> in the way
>>>>>>> proposed.
>>>>>>> 
>>>>>>> 
>>>>>>> On 26 Nov 2015, at 18:47, Martynas Jusevičius <martynas@graphity.org>
>>>>>>> wrote:
>>>>>>> 
>>>>>>> Can't you reuse RDF/POST for that?
>>>>>>> http://www.lsrn.org/semweb/rdfpost.html
>>>>>>> 
>>>>>>> On Thu, 26 Nov 2015 at 19:38, henry.story@bblfish.net
>>>>>>> <henry.story@bblfish.net> wrote:
>>>>>>>> 
>>>>>>>> Hi,
>>>>>>>> 
>>>>>>>> following a discussion with the IndieWeb folks on a pingback
>>>>>>>> protocol,
>>>>>>>> where we are trying to satisfy the constraints that the IndieWeb folks
>>>>>>>> live
>>>>>>>> by - which is very close to the first generation of the Web - and yet
>>>>>>>> who
>>>>>>>> want to work in a global web of data - the following requirement came
>>>>>>>> up: how can a resource make explicit its interpretation of web forms.
>>>>>>>> 
>>>>>>>> I write this up here:
>>>>>>>> https://github.com/w3c-social/webmention/issues/11
>>>>>>>> 
>>>>>>>> Essentially this would not entail very much other than creating a Link
>>>>>>>> realation
>>>>>>>> that could point to something like a document that would contain a
>>>>>>>> mapping
>>>>>>>> that could look like this
>>>>>>>> 
>>>>>>>> [[
>>>>>>>> @prefix foaf: <http://xmlns.com/foaf/0.1/>.
>>>>>>>> @prefix contact: <http://www.w3.org/2000/10/swap/pim/contact#> .
>>>>>>>> 
>>>>>>>> CONSTRUCT {
>>>>>>>> []  foaf:name ?name;
>>>>>>>>  foaf:age ?age;
>>>>>>>>  contact:home
>>>>>>>>          [ a contact:ContactLocation;
>>>>>>>>              contact:address [ contact:city ?city;
>>>>>>>>                                contact:country ?country;
>>>>>>>>                                contact:postalCode ?zip;
>>>>>>>>                                contact:street ?street
>>>>>>>>                              ] .
>>>>>>>> } WITH ?name ?age ?city ?country ?zip ?street
>>>>>>>> ]]
>>>>>>>> 
>>>>>>>> where ?name ?age ?city etc would be the values of the properties taken
>>>>>>>> from the
>>>>>>>> application/x-www-form-urlencoded content sent in a POST.
>>>>>>>> 
>>>>>>>> This should not be that much work to specify I suppose, especially for
>>>>>>>> the
>>>>>>>> SPARQL knowledgeable folks.
>>>>>>>> 
>>>>>>>> Any feedback?
>>>>>>>> 
>>>>>>>>      Henry
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>> 
>>>>> 
>>>> 
>>

Received on Friday, 27 November 2015 15:52:03 UTC