Imagine that a citation resolves to a composite object (e.g. OAI-ORE) via a service. I can discover the components of the object and then pull specific elements down and process them in some defined way. For example: From Paper X I want the structure from fig 5 and I want to see it in this orientiation; or From Paper Y I want video supplementary information and I want to see the frame at 1:45. - Cameron Neylon

p.s. Would this make citations RESTful or am I getting the wrong end of the stick on that? - Cameron Neylon

I guess they'd be RESTful if each component were stored with the appropriate associations and accessible via URLs such as: documents/1/figures/2. document :has_many => [:figures, :tables], figure :belongs_to => document and so on, would be a "Rails-ish" way. - Neil Saunders

@Cameron, (not sure, I didn't read the full spec) but I think that it could be the role, at least for the hyperlinks, of XLINK ( http://www.w3.org/TR/xlink/ ) e.g. it contains the 'from' and 'to' attributes. I don't think it can be translated into a RESTful URL. - Pierre Lindenbaum

I think it would be important to maintain a URL as a citation. I can see that DOIs presumably would break such a thing (just try adding extra text to a doi and send it to a resolver) but that could presumably be changed. Not sure I understand the xlink documentation :-) - Cameron Neylon

Such a system would need a well-developed "you want me to do WHAT with WHAT? are you nuts?" error-resolving technique. - D0r0th34

Agreed. I think I'm making the assumption that the object gets served through an appropriate service (hence media type is important, hence REST, which as I understand it leaves the actual display problem with the client doing the requesting) which can hopefully handle the "I'm not squeezing your foo through my bar" problem. At one level I'm actually more interested in the philosophical question though - what is a citation and what does it _do_. I presume there is a proper literature on the functionality and epistemology(?) of citation? - Cameron Neylon

great question, Cameron, "what is a citation and what does it _do_" - it brings into focus what seemed to be lingering here http://ff.im/7y8SA when Christina said: "I guess the question is: how much is in the citing context and how much is in the citation?" and I suggested "are we sure there really is no concept or usage in between these two?" or something along these lines - or actually we're dealing with hitherto unimaginable lines? :-) - Claudia Koltzenburg

or maybe another way of asking the same question is do we want citations to just point or do we want them to have semantic weight? A lot of this seems to raise some of the same issues as trying to assign "qualities" to citation. Is it positive or negative etc? - Cameron Neylon

Weeellll, seems to me you're also asking whether a link is just a citation... or something more. ;) - D0r0th34

which brings me to asking: when was it invented, in which context, for what purpose, to function how? to what effect for whom? Maybe a 'citation' is a left-over like the print analogy we see in pdfs? - Claudia Koltzenburg

Good question - can anyone point us to a good history of citation? - Cameron Neylon

@D or is a citation just a link or something more? :-) My guess would be that a link is a technical device whereas a citation is intended to have meaning via its context. I guess paraphrasing Christina's comment into my question, can we parcel (some of) that context into a machine readable packet within the citation/link - Cameron Neylon

Would be cool if someone adds "semantic indexing" of scientific papers. That could enable such an API/query-language for all papers which uphold to some conventions (some of which may actually be well established in some scientific communities). One major obstacle could be the PDF format though. Such software would be far more realistic if papers were published in somewhat clean (X)HTML. - Meryn Stol

@Meryn "semantic indexing" of scientific papers : isn't it what is what NLM/MESH is supposed to do ? OK, most of the papers are awfully annotated. Adobe has also created XMP http://www.adobe.com/product... , a technology to add semantics in the pdf. But I don't know any publisher using it. - Pierre Lindenbaum

Wrt. the URL citation: anybody know off hand if this is something OpenURL does (or should) support? so, you could perhaps embed a DOI for a publication (paper, video) in an OpenURL and add parameters to indicate a particular section and whatnot. - 'Mummi' Thorisson

BTW on a related note; some journals register sub- or child-DOIs, so-called components (http://www.crossref.org/help... ) for figures and tables, so presumably these can be cited directly if desired. - 'Mummi' Thorisson

Great discussion, Cameron. I would think about it the other way, though. A citation is clearly a special class of link, and all links can now have some semantic baggage, so there's no reason they shouldn't. - Mr. Gunn

catching up: is there a theory of citation - why do people cite, history of citation etc.... yes, several, http://christinaslibraryrant.b... but there are other articles - Christina Pikas

I'm not entirely sure my brain is up to social theory at the moment...anyone want to fund me to do another PhD or three on this stuff? :-) - Cameron Neylon

isn't part of the point that I do the social science stuff, you do the science-science stuff, we both do the web 2.0 stuff and we make the world a better place? :) - Christina Pikas

Well possibly - but I'm finding the interface stuff much more fun and can't help but feel that a better understanding of the theoretical framework would help. That or I need to find some other excuse to put the time in to learn it :-) - Cameron Neylon

excellent & interesting, how about doing a game of collecting examples of citations (from all over the web) and see which ones we wouldn't have thought of - or which we would not have called 'citations' ;-) ... any web art geeks around? - Claudia Koltzenburg

Oooh - I like it. Maybe we could get Wellcome to fund something on this - they like art-science projects... - Cameron Neylon

brilliant - and indeed this might turn out to be very inspiring for developers and many others, Cameron - Claudia Koltzenburg

So question - do we know any science oriented digital media artists who would be into something like this? - Cameron Neylon

"science oriented digital media artists" The name Gerd Leonhard springs to mind as someone who might be able to assist http://www.mediafuturist.com/ - Graham Steel

seems to me Noomi Ljundell has a hunch for metadata magic :-) http://www.helsinkischool.fi/helsink... - Claudia Koltzenburg

hi all, re "see the frame at 1:45" and other observations, where has this discussion taken you? is anyone approaching the Wellcome Trust yet? - Claudia Koltzenburg

RT @cameronneylon do we know any #science oriented digital #media #artists who would be into something like this? - Daniel Mietchen

To get an overview of the context in which citations of the same item may occur, services like http://brainmaps.org/index... may be of interest. - Daniel Mietchen

An aspect of REST that's still not well appreciated: resource design. All kinds of things can be modeled as resources, in ways we don't normally expect. I have two favorite examples. First, modeling a bank transaction as a resource (see http://blog.jonudell.net/2007...). Second, modeling recent user activity as a resource: http://roy.gbiv.com/untangl.... So, there is surely much creative headroom for modeling a citation as a resource. - Jon Udell

Nice, got some expert input here from Mr. Udell. - Mr. Gunn

thx, Jon, what if the resource should be able to keep on moving, turning, twisting, sounding... whatever? - Claudia Koltzenburg

Interesting. So if I get what Jon is saying then we could imagine a process (as a transaction) and address a process on an object via that. That makes no sense, let me try and make sense of it. We first hit a citation http://journal.org/ref-for.... The REST service returns a list of objects we might wish to deal with - sections of text, images, videos, structures, database entries, whatever, and when we ask for one of those resources we are passed both the resource and a suggested list of services for dealing with it. Because its resful you can either work your way through without initial knowledge of whats available - or if you know what you want you can construct a citation such as http://journal.org/ref-to-... - Cameron Neylon

@Cameron - I don't see why you couldn't have query parameters in there as well, tacked onto the URL for the resource or sub-resource itself: say http://journal.org/ref-for... params specifying a particular part of the figure], or for a video resource http://dx.doi.org/10... specifying the first 3 minutes]. Or a simpler example: http://dx.doi.org/10... to point to a section. - 'Mummi' Thorisson

As Jon said, figuring out what should be modelled as a resource is quite important. For instance, a section of text in a paper to me wouldn't necessarily be a resource in its own right. But a figure in a paper would, as long as the parent->child relationship is made clear. - 'Mummi' Thorisson

could we also do without such hierarchies? have the non-text resources count for themselves? - Claudia Koltzenburg

I wouldn't worry about implied hierachies - I think in effect all the objects become equal but related - the "paper" is just a container that describes a set of relationships between figures, text, and whatever else. Mummi, nothing wrong with query parameters but this is implied by my scheme I think. If you know what you want you just construct the appropriate citation - if you know what you want but aren't sure how to display it you got the object and ask for the service options - if you just want to see what is there you hit the top object. - Cameron Neylon

re set of relationships, Cameron: ok, what I am suggesting is this: to do away with habits inherited from the print world as best we can in order to open up more creative space :-) - Claudia Koltzenburg

Ah ok - I'm just thinking more of making things technically possible. So if as a first naive step we wanted to create a new paper out of pre-existing pieces we would generate a new container that pointed at all the pre-existing pieces but that is kind of obvious. Something more interesting...I want to invert the conclusions of a published body of work by drawing attention to a specific way of viewing a related data set? How do I do that? Hmmm - Cameron Neylon

Related: RDFa of citation ontology, embedded in http://inkdroid.org/journal... - Jodi Schneider

reminds me some of Catherine Blake's work:http://people.lis.illinois.edu/~clblak... but I'm having trouble putting my finger on what I'm thinking of. - Christina Pikas