- From: Steve Harris <steve.harris@garlik.com>
- Date: Fri, 25 May 2012 10:30:19 +0100
- To: Manu Sporny <msporny@digitalbazaar.com>
- Cc: RDF Comments <public-rdf-comments@w3.org>, RDF WG <public-rdf-wg@w3.org>, Linked JSON <public-linked-json@w3.org>
On 2012-05-25, at 03:47, Manu Sporny wrote: ... >> Section 3.1, line 11: "A value is an object with a label that is not >> an IRI" >> >> This is not technically wrong, but I would urge you in the strongest >> possible terms to reconsider using the term "value" for a graph node >> of any kind. > > This terminology came out of a very long series of discussions where > folks repeatedly stated that the RDF terminology was confusing and > overly pedantic. That said, we can word smith this to try and find a > common ground... > > For example - the word "predicate" results in confused looks from Web > developers where the word "property", while less accurate, does not. > Similarly, "literal" is harder for beginner developers to understand > than "value". I agree that generally the wording of the specs for RDF has been tilted very far towards readability for people familiar with formal logic, at the expense of comprehensibility the developer on the street. OTOH I don't think that *users* of technologies should necessarily have to read the specs, but as a community we've been bad at writing tutorials (when when we do those tutorials often use the confusing* terminology of the logicians too), so often they have little choice. Sadly, the longer you spend in these circles the more you forget how antisocial this terminology is. * confusing for the lay-developer - e.g. while the edges in RDF graphs are technically predicates, that's both not the sense of the word that many developers are familiar with, and not relevant. >> Finally, a pet peeve of mine regarding blank nodes. >> >> [LD] 3.1 says >> >> " Unlabeled nodes are not considered Linked Data." >> >> Says who? I didnt know there was a Ministry (Church?) of Linked >> Data. > > Based on the definition of Linked Data that the group found consensus on: > > http://json-ld.org/requirements/latest/#linked-data > > Specifically: > > "An IRI that is a label in a linked data graph should be dereferencable > to a Linked Data document describing the labeled subject, object or > property." > > A blank node doesn't allow you to create links between documents and > since you can't create links, it doesn't fit the definition of Linked > Data that we have consensus on. However, we did find that it does > constitute Structured Data (also defined in the Requirements document). This is quite a religious discussion. I think that, pragmatically, discouraging people from using bNodes with this kind of wording is a bad idea. For e.g. if I have <london> a :Place; :centre [ geo:lat 51.507977; geo:long -0.124588; ]. There's no good reason to give the actual coordinates a globally dereferencable URI, it's more work for everyone for no gain. I'd have to go to the trouble of minting a stable URI for it for e.g., and face some tricky decisions if the centrum of London were to move, for example. bNode genid URIs give us a way to get the system to mint a URI, taking the strain off the developers, while still being able to use it in SPARQL results and so on. I'm certain that not everyone agrees with me, just pointing out it's not a widely held definition of Linked Data. - Steve -- Steve Harris, CTO Garlik, a part of Experian 1-3 Halford Road, Richmond, TW10 6AW, UK +44 20 8439 8203 http://www.garlik.com/ Registered in England and Wales 653331 VAT # 887 1335 93 Registered office: Landmark House, Experian Way, Nottingham, Notts, NG80 1ZZ
Received on Friday, 25 May 2012 09:30:59 UTC