It’s been years and years since I did either a formal literature review, or used a reference manager like EndNote or RefWorks in anger, but whilst at the Arcadia Project review in Cambridge a couple of days ago, I started wondering what sorts of ‘added value’ features I’d like to see, maybe even expect, from referencing software nowadays…
One of the ideas I’ve been playing with recently is the idea of emergent social positioning (ESP;-) in online social networks, which I’m defining in terms of where an individual or an expression of a particular interest group might be positioned in terms of the socially projected interests of people following that person or interest group.
For the case of an individual, the approach I’m taking is to look at who the followers of that individual follow to any great extent; for the case of an interest group, as evidenced by users of a particular hashtag, for example, it might be to look at who the followers of the users of the hashtag also follow in significant numbers.
A slightly more constrained approach might be to look at how the followers of the individual or the hashtag users follow each other (a depth 1.5 follower network about an indvidual or set of individals, in effect).
So for example, here’s a map I just grabbed of folk who are followed by 3 or more followers from a sampling of the followers recent users of the #gdslaunch (Government Digital Service launch) hashtag.
So what does this have to do with reference managers? Let’s start with a single academic paper (the ‘target’ paper), that contains a list of references to other works. If we can easily grab the reference lists from all those works, we can generate a depth 1.5 reference map that show how the works referenced in the first paper reference each other. Exploring the structural properties of this map may help us better understand the support basis for the ideas covered in our target paper.
By looking at the depth 2 reference network (that is, the network that shows references included in the target paper, and all their references), we may be able to discover additional (re)sources relevant to the target paper.
Unfortunately, getting free and and easy machine readable access to the lists of references contained within journal articles, conference papers and books is not trivial. There are patchy services such as CiteSeer, Citebase or opencitations.net, but I don’t think services like Mendeley, Zotero or CiteUlike are yet expressing this sort of data? Or maybe they are, and I’m missing a trick somewhere.
(Just by the by, presumably some of the commercial citation services have APIs that support at least accessing this data? If you know of any, could you add a link in the comments please?:-)
Another hack I’d like to try is to generate what more closely corresponds to the social positioning idea, which is to grab the references from a target paper, and then the papers that cite those references and see how they all link together. This would help position the target paper in the space of other papers referencing similar works. I think CiteSeer has this sort of functionality, though not in a graphical form?
PS on my to do list is seeing whether I can get reference lists for articles out of Citeseer using the Citeseer OAI-PMH endpoint. I’ve got as far as installing the pyoai Python library, but not had time to try it out yet. If anyone knows of a guide to OAI for complete novices, ideally with pyoai examples I can crib from, please post a link (or some examples) via the comments:-)