Knowledge organization, reading notes, killer apps
At the end of an article on theory in knowledge organization that I
read for my cataloging class, I found (and circled) this paragraph:
…Another
area rife [sic] for theoretical development is the extensive work of
cocitation and coword analysis. This work describes relationships among
scholars, essentially mapping intellectual relationships within
knowledge domains as represented by citations and abstracts. What is
needed are sociological (i.e., cognitive) explanations of the behaviors that lead to these
intellectual relationships. Such explanations could give us real
predictive power for the development of sophisticated systems for the
retrieval of knowledge entities. (Richard P. Smiraglia, "The progress
of theory in knowledge organization," Library Trends 50 no. 3 [Winter
2002], p. 346)
To paraphrase a bit: you could create some potentially really
useful research tools using the data you get from analyzing who’s
citing what, which articles refer to each other, which works always
appear together in Works Cited lists, and how scholars in a field are
interrelated. It would be helpful to have some data on scholars’
behavior, to describe why these citation patterns happen.
I don’t have a cognitive science background, or a sociological
one. But this kind of thing — figuring out where the intellectual
terrain of a field meets the social networks of the people in it, using
patterns in the scholarly literature to (potentially) develop the next
killer app for navigating the intellectual output of that field —
interests me. A lot. It would be very cool to be the person who mediates between the scholars and the information researchers. If I ever get the chance to work on a study like
this, I’ll jump at it.
And while I’m thinking about killer apps: ISI Web of Science,
clunky and flawed though it is, is a fairly powerful
citation searching tool once you know how to use it. But what I’d love
to see from a citation search, beyond a list of articles that all cite
a certain source, would be groupings showing which articles cluster
together (X, Y, and Z all cite A and B; X cites C, which D and E also
cite; D’s work shows up in Y’s recent article; etc.). Web of Science
does this to some extent, but in an ideal world, I’d also like to have
the clusters presented visually, Grokker-style, so you could see the relative density
of interconnections and the offshoots into other disciplines. Maybe
even make it possible to see how the conversation evolves over time.
And since it would take a gazillion years to produce all of
this citation data, I also think it would be neat (though this is a
total pie-in-the-sky goal) to get scholars themselves involved. Not
with the programming and indexing, but with identifying where the
linkages are.
The question is: where on earth can one start with a project
like that? I suspect this is a question I’ll have when I take Content
Representation in the fall.
From the sociological/organizational perspective, you want to go back and look at work on “invisible colleges”. Especially Derek J de Solla Price and Diana Crane. I was surprised not to see a reference to this in the Smiraglia abstract in your link.
I used to work for a group that did information retrieval and text mining (mostly on news and biomed corpora). One of the guys in that group built a tool that would perform a search and display the results of the search as a social network. It was not exactly speedy, but it was a useful way to display and quantify connections that otherwise weren’t readily obvious.
I *knew* there was research along those lines out there — thanks, oliviacw! (I’m also surprised the article didn’t mention it, but then again, the bit that caught my eye was kind of tangential to the main topic.)
Withneedle, that sounds very much like what I’m thinking of — I don’t suppose there were any plans to build a version of that tool for more general use?
Sorry for the slow response. There weren’t ever plans that I know about to build a general purpose version of the social network viewer. It would have likely have been a pretty big project to get the tool to perform well enough to be generally useful. However, that group as taken its technology to a startup, so all kinds of things could happen there…