Re: Content negotiation negotiation

On 4/24/13 7:06 AM, Hugh Glaser wrote:
> On 23 Apr 2013, at 22:39, Kingsley Idehen <kidehen@openlinksw.com>
>   wrote:
>
>> On 4/23/13 5:04 PM, Hugh Glaser wrote:
>>> Ah POWDER - of course.
>>> It all comes together :-)
>>> (Sorry if this is boring and obvious to others - and thanks Kingsley.)
>>> So last (?!) 2 things, if I may.
>>> Any proposal to attach types to the objects of the wdrs:desribedby triples?
>> So as in <http://ns.nature.com/docs/terms/datatypes/anyURI___279277607.html> you seek the xsd:anyURI type qualification, for objects of said relation, right? If yes, then fine, it can be added quickly.
> I don't think so.
> I am not after a datatype.
> In fact I find datatypes particularly unhelpful in an RDF/SPARQL context, but that's another story.
>
> What I am after is the MIME type of the different alternates that are offered for conneg.

I think you seek:

1. triples for xhv:alternate relations
2. triples for content-type relations.

Not an issue to add at all.
> So I can query something like
> SELECT ?source FROM { <http://dbpedia.org/resource/Luton> ?foo ?file . ?file mime:application/json ?source . }
> and then choose my json source.

The items above facilitate that.

>
> All that wdrs:described by tells me is that there is another file, but nothing about the format.
> (Of course the "json" at the end is a hint, but it is really just an opaque string.)
> I could of course resolve them all, asking for json or whatever, and see what the response code/header gives me back, but I don't want to/can't really see headers in the Linked Data context. And the server provider would not thank me for the traffic.
> And in any case, many servers are particularly bad at giving a 406 or whatever it is, and simply give a 200 and html.
>
>>> Any proposal so that I can infer the available types for the whole dataset, rather than inferring from a particular resource resolution?
>> You mean for RDF resources such as the one denoted by <http://dbpedia.org/data/Luton.ttl> ? If yes, then we can just add the missing resource metadata relations which would basically come from VoID [1].
> I mean for the whole dbpedia.org/resource dataset.

So you need a VoiD graph and SPARQL endpoint description graph. Both 
should exist, so I just need to check why that are missing or not 
exposed via RDF.

> I am guessing that the range of content for most significant sites (such as dbpedia) is actually the same for all resources.
> So I could build a KB that had the formats available.
> In fact, I might even add it to my voiD store.

As per comments above :-)


Kingsley
>> Links:
>>
>> 1. http://www.w3.org/TR/void/#class-property-partitions
> By the way, I can't seem to see the wdrs:describedby stuff in http://dbpedia.org/resource/Luton at the moment.
>
> Best
> Hugh
>> Kingsley
>>> Cheers
>>>
>>> On 23 Apr 2013, at 21:48, Kingsley Idehen <kidehen@openlinksw.com> wrote:
>>>
>>>> On 4/23/13 4:23 PM, Hugh Glaser wrote:
>>>>> Ah, thanks for the Web101 course.:-)
>>>>> Sorry, I usually live in a Linked Data world, so I don't think about html stuff such as
>>>>> <link rel="alternate" …
>>>>> because (like the header) it doesn't appear in the RDF.
>>>>>
>>>>> On 23 Apr 2013, at 20:54, Kingsley Idehen<kidehen@openlinksw.com>
>>>>>   wrote:
>>>>>
>>>>>>> On 4/23/13 3:39 PM, Hugh Glaser wrote:
>>>>>>>>> Ah of course - thanks Mark, silly me.
>>>>>>>>> So I look at the Link: header for something like
>>>>>>>>> curl -L -ihttp://dbpedia.org/resource/Luton
>>>>>>>>> Which gives me the information I want.
>>>>>>>>>
>>>>>>>>> Anyone got any offers for how I would use Linked Data to get this into my RDF store?
>>>>>>> Assuming I understand your question, the answer would depend on the capabilities of your RDF store. If it can injest RDF resource URLs you can request the formats exposed on the "Link:" responses.  If it handles SPARQL 1.1 INSERT and/or LOAD just use SPARQL.
>>>>> I don't think I can use the SPARQL INSERT, etc, because it isn't RDF.
>>>>> Is the <link rel="alternate" available anywhere as RDF?
>>>>> It could be returned with the RDF forhttp://dbpedia.org/resource/Luton  Better still, it could be available in the voiD description (so that it is site-oriented, not resource-oriented)?
>>>>> Or somewhere else?
>>>>> Cheers
>>>> Okay, now that <link/>, "Link:", and SPARQL aren't options, of course you can get it from the RDF that describes <http://dbpedia.org/resource/Luton>, see:
>>>> http://dbpedia.org/describe/?url=http%3A%2F%2Fdbpedia.org%2Fresource%2FLuton&gp=8&go=
>>>>
>>>> We use the wdrs:desribedby relation for that :-)
>>>>
>>>>
>>>>
>>>> -- 
>>>>
>>>> Regards,
>>>>
>>>> Kingsley Idehen	
>>>> Founder & CEO
>>>> OpenLink Software
>>>> Company Web: http://www.openlinksw.com
>>>> Personal Weblog: http://www.openlinksw.com/blog/~kidehen
>>>> Twitter/Identi.ca handle: @kidehen
>>>> Google+ Profile: https://plus.google.com/112399767740508618350/about
>>>> LinkedIn Profile: http://www.linkedin.com/in/kidehen
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>
>> -- 
>>
>> Regards,
>>
>> Kingsley Idehen	
>> Founder & CEO
>> OpenLink Software
>> Company Web: http://www.openlinksw.com
>> Personal Weblog: http://www.openlinksw.com/blog/~kidehen
>> Twitter/Identi.ca handle: @kidehen
>> Google+ Profile: https://plus.google.com/112399767740508618350/about
>> LinkedIn Profile: http://www.linkedin.com/in/kidehen
>>
>>
>>
>>
>>
>
>


-- 

Regards,

Kingsley Idehen	
Founder & CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen