Re: adding hypermedia to spatial data best practices

hello frans.

On 2015-08-05 4:24 , Frans Knibbe wrote:
> The way I understand it, RDF is not a data format but a data model, a
> way of structuring data and modelling the world. It is something like
> the relational model that is used in relational databases.

yes, i probably should have said model and not format. but one way or 
the other, it's simply a specific way to model graph data, and in itself 
has no relation to hypermedia models/formats.

> It depends on who you ask, but ties between Linked Data and RDF need not
> be that strong. The original principles
> <http://www.w3.org/DesignIssues/LinkedData.html> mention RDF as one of
> 'the standards' that should be used but the Five Star data schem
> <http://5stardata.info/>e does not mandate RDF for any level of stars.

yes, it depends on who you ask. but most people assume that linked data 
*is* RDF by definition. quite a while ago, this happened:

http://dret.typepad.com/dretblog/2009/11/the-linked-data-police.html

since then, i have been careful to not use the term when talking about 
the more general concept of data that is linked. and just this past may, 
i was at the LDOW workshop at WWW2015 and once again encouraged the core 
community to be less dogmatic about *what* data model to use, and to 
simply focus on the linking. but most people did not like the idea and 
preferred to stay within the world view that in order for something to 
be linked data, it must be RDF.

that's why https://github.com/dret/webdata came into being. i think 
there's huge value in linking data, and RDF is just one way to do it. if 
people find another model a better fit, they still should get some 
guidance on how to link it in the best way. web data is supposed to be 
linked data without the "must be RDF" constraint, and i think it could 
be useful to have a term (maybe "web data", maybe something else) that 
would be broader than one specific technology, but still focus on the 
linkage of data that creates so much value when used properly.

> It just so happens that RDF is a model that is very well suited for data
> on the web: you can model just about anything and data elements are
> identifiable by HTTP URIs.

sure. i don't dislike RDF. it has many advantages. but like any 
metamodel, it's a good fit for some things and not such a great fit for 
others. you can also model just about anything in XML, JSON, SGML, or 
any other metamodel language. people exposing data should have the 
freedom to pick the metamodel that's the best fit for their requirements.

> I sincerely hope that we can work without a strict distinction between
> data on the web from a REST/human perspective and a Linked Data/machine
> perspective. I hope we can satisfy the needs of both worlds in a single,
> well considered pass.

i agree, and i think that's why ideally, there should be no dogmatism 
when it comes to choosing technologies. what matters most in creating 
the value of webby data is the pattern, and for the web, that pattern is 
hypermedia. (or what REST calls HATEOAS, easily earning the price for 
the worst acronym ever.)

cheers,

dret.

-- 
erik wilde | mailto:dret@berkeley.edu  -  tel:+1-510-2061079 |
            | UC Berkeley  -  School of Information (ISchool) |
            | http://dret.net/netdret http://twitter.com/dret |

Received on Wednesday, 5 August 2015 16:40:35 UTC