Talk:Resource Description Framework

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Computing  
WikiProject icon This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
 ???  This article has not yet received a rating on the project's quality scale.
 ???  This article has not yet received a rating on the project's importance scale.
 

Self-reference[edit]

Per Avoid self-references, should this article really link to an RDF dump of Wikipedia's contents (under the "RDF Files" section)? – Minh Nguyễn (talk, contribs) 04:20, 27 July 2006 (UTC)

It is an example of self-reference, but much milder than the examples given in Wikipedia:Avoid self-references. Its removal would not detract much. If you feel bothered by it, remove it. cygri 20:12, 11 February 2007 (UTC)
I am removing it. However, Example 2 is more serious self-reference problem and should be removed if someone can write a good second example (I don't have the technical knowledge of RDF to do it); the first simple one is not enough. However, there is a world of examples to choose from, which makes a WP article a particularly poor choice. One reason for the Wikipedia:Avoid self-references policy is that it makes WP look amateurish and therefore less authoritative. EB doesn't do it in the many articles where it could use itself as an example. Finell (Talk) 21:33, 17 February 2007 (UTC)
Is there an external reference that uses Wikipedia as RDF somehow? Drf5n 15:48, 2 November 2007 (UTC)

Planet RDF[edit]

I added a link to planetrdf.com since that site really is the hub of all RDF-related development. Subsequent edit took it out citing policy against linking to blogs. I'd contest that - planetrdf.com is an aggregator for RDF-related news and is central to the RDF community. Omitting it seems detrimental to this article. – Ian Davis

The site was described as a blog aggregator, and per Wikipedia:External links blogs are normally to be avoided. If you feel that this link is important, I won't oppose you adding it again, but I would appreciate it if you would explain here in some detail how it is important to the article. That would also help if somepne else comes along and also sees that it is a blog. -- Donald Albury 01:10, 5 September 2006 (UTC)
Here comes someone else in support of Ian, and re-insertion of the link to Planet RDF. RDF community is a reality, and Planet RDF is a hub for this community. I understand why links to blogs should generally be avoided (they often provide non-neutral, biased or extreme viewpoints), but since Planet RDF is aggregating many blogs, this makes for multiplicity of viewpoints. universimmedia 07:36, 5 September 2006 (UTC)
(Disclosure: Ian Davis, universimmedia, and me are all syndicated at Planet RDF.) Is a blog search engine a blog? No. Neither is a blog aggregator. The link should be in the article. Planet RDF is the most substantial source of RDF-related news and includes almost all of the important voices of the RDF community, including several members of the W3C's working groups that have developed RDF. The policy cited above explicitly does not apply to sources “written by a recognized authority”. Since Planet RDF is an aggregator of more than 50 sources, it is very different in nature from a single-author blog, and I do not see how the cited policy applies. (I leave it to someone else to re-add the link because of the obvious conflict of interest.) cygri 20:07, 11 February 2007 (UTC)

RDF data model vs. RDF/XML syntax[edit]

The article should discuss the difference between the RDF data model (a.k.a RDF Abstract Syntax) and the RDF/XML syntax. As it stands, the article mostly discusses the data model, but the file format infobox refers to the RDF/XML syntax, and the term RDF/XML is used throughout the article without the necessary introduction. cygri 09:05, 13 March 2007 (UTC)

Can't Add References?[edit]

I've attempted to add a reference in the Criticism of RDF section but the wiki is not creating a reference section. Can anyone fix this?Clan-destine 23:53, 11 June 2007 (UTC)

The filling-in of the references section is automatic, but you still have to create a stub for it. I just created one by adding the following to the article, above the "See also" section:
==References==
{{subst:Reflist}}
The Reflist template fills in the actual code, the crucial bit of which is <references/>mjb 02:31, 12 June 2007 (UTC)

other examples[edit]

I'm new to RDF, but it seems this is another way of expressing a function based relationship. Given the article's example:

<http://en.wikipedia.org/wiki/Tony_Benn> <http://purl.org/dc/elements/1.1/publisher> "Wikipedia"

it could be expressed as y = f(x):

"Wikipedia" = http://purl.org/dc/elements/1.1/publisher( http://en.wikipedia.org/wiki/Tony_Benn )

and since y = f(x) is commonly expressed in various ways in computer code, wouldn't an example like this be useful in helping convert specialized jargon to a mathematical-like language that transcends written language, such as English?


anonymous-pion —Preceding unsigned comment added by 71.237.190.111 (talk) 12:00, 15 September 2007 (UTC)

I may be risking WP:NOTFORUM here, but as this is a good faith suggestion for article improvement (albeit an old one), I'll respond to that.
The problem with this is that a triple statement of that form doesn't map onto y = f(x). It is possible to have multiple statements with the same subject and predicate but a different object. Consider the Dublin Core predicate for author: it is perfectly reasonable to wish to say a book has multiple authors. It is therefore not a mathematical function: the predicate and the subject for both statements is the same, and so per the principle that a function gives an identical result with the same argument, it's not a function.
You could rewrite the formulaic expression such that the set of all statements within a world can be quantified over and selected by a function f(x) such that the function represents the predicate and the first argument represents the subject if you represent the value of that function as, say, an unordered set of values. Those values would obviously have to represent the RDF semantics, so you'd have to cope with plain literals (like the word "Wikipedia"), URIs (like http://en.wikipedia.org/), language literals (the word "Wikipedia" with the accompanying language tag "en" for English) and datatype literals (the number 42 plus a datatype annotation URL to note that it is an integer).
Reducing that to the language of mathematics might be rather confusing (don't believe me? Read the W3C specs.) but is certainly doable.
It won't help you much because in practice, there's the practical issue of gathering RDF statements and inferring from them. The statements that match a subject-predicate pair in an open world assumption won't necessarily match in a situation of closed world assumption plus inference. Take the following:
Ship A is in California. Ship B is in Virginia. Ship C is somewhere, but we don't know where.
When you take those statements, switch to a closed world assumption, and run inference over them, given an inference rule of "We have one ship in California. Ships are either in California or Virginia." the extension of the predicates and subject-predicate pairs is going to differ. Specifically, the location predicate and the Ship C/location subject-predicate pair will change their extension.
I think RDF isn't necessarily the easiest thing to understand. But explaining RDF in terms of the formal mathematical representation just means you go from having one confusing syntax that's tailor-made for the problem to another syntax that's not. "Now you have two problems" as hackers like to say. —Tom Morris (talk) 17:05, 7 February 2012 (UTC)

Literals in subject position[edit]

RDF does not allow literals to occur in subject position. Therefore the very first example of this article stating that "the sky" is blue cannot be expressed in this way. Instead one would have to associate a URI with the concept "sky". I don't want to mess up the introduction, but this should be corrected. Brezenbene 16:31, 2 November 2007 (UTC)

Well it says a 'subject denoting "the sky" '... so it is not false, actually. But a little confusing, because one might think of literals. I guess we can delete my comment above. Brezenbene 16:34, 2 November 2007 (UTC)

No, I think your comment is very important. The article says that subjects must be URIs but there is absolutely no discussion of: "What is the URI for 'sky'?". This must be very confusing for any newcomer. I think the (sky, is, blue) example should be elaborated, and written in full with URIs. I have no idea what the URI for 'sky' would be, but if I want to actually express that the sky is blue, how would I then write it? If the exmample is too ridiculous maybe another example should be used, e.g.: "Berlin is in Germany" which must be smth that somebody would want to express. 85.101.129.75 (talk) 22:15, 12 January 2012 (UTC)

http://dbpedia.org/resource/Sky seems like a pretty good URI for the concept of the sky. Face-smile.svgTom Morris (talk) 17:07, 7 February 2012 (UTC)

Added Notes and rearranged references[edit]

I added a Notes section to gather footnotes and moved the W3C links into the references section. StephenReed 17:05, 26 March 2008 (UTC)

Microsoft[edit]

Another article claimed that Microsoft contributed to the initial development, but no reference was cited. Is this true? -- Beland (talk) 05:18, 16 October 2008 (UTC)

Criticisms?[edit]

Hi, I'd like to see the criticisms section made more substantial. Expanding it into narrative form and going into a bit of depth for each criticism/counter-criticism would be highly educational and enrich this article's value in terms of encyclopedia content. I'm not qualified to do this (I don't know much about RDF) but I do not that there must be additional criticisms ("Are billions of tuples a feature or a bug?", authenticity, ambiguity, provenance, etc.) that would be helpful to include here. -- Joebeone (Talk) 12:35, 7 August 2009 (UTC)

The criticism section is essential and needs considerable expansion. I'm not concerned with the issues of verbosity, but I'm involved in concrete applications where provenance, authentication, security, access are required and complex. A single triple expands into perhaps dozens of statements about its reification. RDFa supports all of this, but support is rare among tools and libraries - maybe even non-existent. Without this support, tracking of provenance is, in theory, possible, but everyone would have to do it "by hand", with the ensuing chaos and duplication of effort. -- Deeptext (Talk) 17:35, 20 August 2009 (UTC)

Today I deleted the entire section. The one reference it had is no longer functioning and was not archived at the Internet Archive, as far as I can tell. Recently added material about 'reinvention of the wheel' and PROLOG's support for triples really isn't criticism, since RDF wasn't claiming to have invented or perfected the semantic triple, and RDF and PROLOG aren't directly comparable.
Furthermore, to address some of your points above, the fact that RDF is not the right tool for every knowledge representation job, or is not a complete knowledge representation system unto itself, just means RDF is no different than any other technology. Some tech is good at some things, not so good at others. It's like "criticisms of toothbrushes: they don't get between teeth as well as dental floss, and they're just reinventions of the toothpick, and to get my teeth really clean I have to go to the dentist" — true statements, perhaps, but ultimately a comparison of apples and oranges, not a valid criticism of the toothbrush, per se.
More importantly, Wikipedia cannot be the first place criticisms are published. If you can find already-published sources for criticisms, you can summarize and cite them here. They should meet Wikipedia's criteria for reliable sources, though, so random blog posts probably aren't going to cut it. —mjb (talk) 23:59, 12 April 2010 (UTC)
I for one would *really* like to see a section on criticism. I'd like to understand whether this is snake oil and hype, or the best thing since sliced bread. As the article stands, I have no idea. Seeing criticisms--and critics--helps me understand that.
For example, fans often talk about how RDF is needed to make sense of the www. I think I can see how it would work if it existed, but have no idea how it could ever come to exist; an example dealing with kinds of Merlot wines certainly don't engender any confidence that this is practical. A criticism that dealt with how RDF triples are supposed to get build on the web scale would help.
But maybe my skepticism in the practicality of this notion is misplaced--a section on criticisms (with answers) would help me understand. Mcswell (talk) 19:53, 11 October 2011 (UTC)
Most of the RDF community have better things to be doing. Many of the amateur-hour alternatives (microformats for one) don't, so they produce more coverage here. It's also why en:RDF still has bogosities like "metadata" in the intro. WP:Notability is a very poor measure of notability.
If you want to see the leading edge and the interesting stuff, take a look at Protege (Stanford) and OWL work. Mind you, one of the more inane comments I've seen on WP was that OWL was being a success instead of RDF. Andy Dingley (talk) 20:12, 11 October 2011 (UTC)
"Most of the RDF community have better things to be doing." Sorry, I don't see what that has to do with the question of (re-)adding a section here on criticisms. I'm not asking the most of the RDF community to respond, I'm asking *someone* to respond so that those of us who are not insiders in the RDF community can read about this. After all, wikipedia is an encyclopedia for the world, not just for the RDF community.
I'm also not asking about leading edge stuff, nor what the RDF community finds interesting. I'm looking for why I--or anyone else not already into RDF--should find RDF to be interesting and practical. Mcswell (talk) 13:27, 14 December 2011 (UTC)
WP has been downright hostile to SemWeb experts so far, going so far as to threaten topic bans on the grounds that anyone who understood a topic thus has an impossible COI and so must be excluded. The obvious result is that the community most able to contribute is instead persuaded to ignore WP. Andy Dingley (talk) 13:55, 14 December 2011 (UTC)
I second Mcswell's desire for a section on criticism. Without it, this article has reduced usefulness in my scenario, which is trying to decide whether I should support it in my project or not. Bjornte (talk) 09:04, 12 February 2015 (UTC)
I would suggest that you make project management decisions on something a little more robust than WP. Andy Dingley (talk) 10:55, 12 February 2015 (UTC)

Applications[edit]

In this section I can read: "Some uses of RDF include research into social networking. This is important because it could help governments keep track of terrorists cells."

Come on!! Seriously?! I'm trying to picture in my mind the Secretary of Defense urging to define a new URI for terrosists, so that they can uniquely described by an URI. Should a terrorist be a FOAF of rather an EOAF (Enemy Of A Friend)? But I think terrorists have friends too, so they might as well be described with FOAF...

Let's write something more verifiable, or in any case more suited to the scope on encyclopedia. Either put a valid reference (but i doubt you can find anybody in the world fighting terrorism by using RDF) or just delete this last sentence.

Squalho (talk) 22:02, 2 November 2009 (UTC)


Some one deleted the terrorism mentioning, maybe you, which is fine. But I want to add smth. Well, I'm not the most insightful, but since nobody else wrote, I will try. If you model all the relationships between citizens, that is just a very powerful piece of information. A big part of the discussions of potential applications of semantic web has been about FOAF, that you can model how people know eachother. If you had all this data you could run algorithms to find suspicious people with suspicious friends. It is a great tool to keep track of any part of the population: minority groups, political opposition, criminals, illegal marriages of conveniance, social fraud, mafia, money laundering, etc. And well also those political groups that some times turn to terrorism. I am absolutely sure that models of interpersonal links, like the Facebook graph, could be used by intelligence services to find e.g. "suspicious acting potential terrorists". And I even think they probably do it to some extent now. Exactly how the intelligence services would manage to put the entire population into a FOAF triple store, I don't know, but I'm sure they would like to, and I'm sure that they are looking into the possibilities 85.101.129.75 (talk) 22:27, 12 January 2012 (UTC)

What is an RDF vocabulary, class and properties?[edit]

I understand the first part of the article quite well, RDF seems in essense to be about (subject, predicate, object) triples.

But out of nowhere, there is a list of Classes and Properties, without any introduction at all. And unfortunately googling does not provide easily understandable answers to this either, but maybe it is not supposed to be easily understood?

Whoever will consider describing these three concepts, please try to do so, assuming that the reader does not already know anything. Unfortunately, imho, all WP articles on semantic technologies are very poor at explaining this to newcomers, but instead just summarize for ppl who already understood.

85.101.129.75 (talk) 23:00, 12 January 2012 (UTC)

The 2004 W3C documents on RDF are still very good, and there's a reasonable primer in there.
As to this article, then yes, it's unreadable. The Classes you see (approximately speaking) refer to instances of resources (or at least, their superclasses) that one might use in the context of subjects & objects for RDF triples. Note that they're capitalised by convention. The Properties (lowercased) have a similar role for the predicates. Note that RDF is very often recursive (see reification), so it's common to want to make statements about other triples; i.e. a triple (the whole tuple of it) can itself be the subject for another triple. Andy Dingley (talk) 00:03, 13 January 2012 (UTC)

Some POV and "textbook" talk in the article?[edit]

I am reading the article and there seem to be some POV opinionated opinions creeping in: "This is important because it could help governments keep track of undesirables." Is it important that it is done with RDF? Is it an opinion that this is what governments will use it for? I find that a highly debatable statement. There are a few other instances of POV creep, and one can tell that this is a pretty "pro-RDF" document.

The article also seems to heavily fall afoul of the Wikipedia is not a Manual policy.

What do others think? --Petercorless (talk) 06:15, 11 July 2012 (UTC)

I recently stumbled across this sentence and I think it is highly opinionated, less related to RDF but rather that it promotes the opinion that "tracking undesirables" is something which is "good". I think this (electronic tracking of people) is a problematic topic and it has as at least as many disadvantages as it has advantages. 10 November 2013 — Preceding unsigned comment added by 84.114.167.240 (talk) 15:30, 10 November 2013 (UTC)

HTML5, RDFa, RDF-Lite and RDF[edit]

This article has no mention or link of HTML pages as resources.

The ongoing changes to HTML5 have removed (or broken ?) a good deal of what was available prior to HTML5 if the current state of invalid Dublin Core is any indication.

Are HTML resources, in fact, such a negligible part of the semantic web or are we seeing the bias of institutional users of web services?

G. Robert Shiplett 00:42, 10 November 2012 (UTC)

Comparison to OO notation[edit]

Quote: For example, one way to represent the notion "The sky has the color blue" in RDF is as the triple: a subject denoting "the sky", a predicate denoting "has", and an object denoting "the color blue". Therefore RDF swaps object for subject that would be used in the classical notation of an entity–attribute–value model within object-oriented design; object (sky), attribute (color) and value (blue).


This makes absolutely no sense to me, I'd even go so far as to say that it is just plain wrong. In RDF you would simply get two triples out of this, not one, with "blue" as a literal in the second triple.

T1: subject "sky", relation "has", object "color" . T2: subject "color", relation "is", object "blue" .

In JAVA you would have class sky { private String color = "blue"; }

the implicit meaning of course being that "the color IS blue" + "the sky HAS A color".

147.142.164.94 (talk) 13:53, 9 April 2015 (UTC)

You could state that in RDF - but it would lose a lot of the benefits of using RDF. Like trying to program by reading Hello Word examples, trivial examples don't explain much.
The advantage of "blue" being an object rather than a literal is that it allows it to have a type (from RDFS or even an OWL ontology) which in turn allows it to have a parent type and it might also have properties in turn.
Imagine that "the Sky is Blue". Blue in this case is an object with a property for "palette". Which could have values of "Dulux paint range" or "Web safe colour palette". Different blues, but now conclusively identified.
The sky might be "Duck Egg" instead. I have no idea what "Duck Egg" or "Spindrift" are [1], but because these are objects in RDF, and they have a type with implicit taxonomy from the schema, I might be able to infer that they're "some sort of blue". If "I" am actually a paint-spraying web agent 'bot with a limited palette, that could be really useful to me. It's even something that's implementable with a few lines of RDF tree-walking code, rather than the language processing and knowledge base needed to work with the literal string "blue".
"The Sky is Blue" might also be vague in the sense that "a poetical statement has been made about the Sky, with the value of Blue". A statement that doesn't even imply colour as a property. Yet because the object Blue is identifiable as a colour, it allows us to sub-type the generalised poetic statement into one about the sky's colour.
So for trivial examples, it doesn't matter because they're trivial. It still doesn't even matter because an object alone isn't much better than a literal. But once you have type associated with objects, all sorts of new things become both possible and simple enough that you can look seriously at implementing them. Andy Dingley (talk) 14:46, 9 April 2015 (UTC)