Thursday, May 12, 2016

Does citation network topology change over time?

A book I am reading makes in passing, and without a reference, a broad sweeping claim that "Internet is killing good research habits". It claims that in the olde times people would go to the library and actually read, while these days they just google, find a paper with most citations, and use it in their work.

It sounds plausible, but it feels like the opposite statement would also sound very plausible. In the old days you would read 3 papers, get used to citations from these papers, and got stuck with them for the rest of your career, while these days you can google-scholar or pubmed for any combination of keywords and find papers from far removed disciplines, institutions, working groups and subfields that you would have never found on your own.

In terms of citation network topology, the first claim implies that in the past the distribution of node orders was more uniform than now, while the second line of thought suggests that actually it might have been more extremely non-uniform (skewed) than now.

I tried to find out the truth by googling about "Citation network evolution" and other stuff like that, but could not find anything. Apparently the distribution was very skewed even in the past, with few papers receiving a status of "classics", and getting thousands of citations. This phenomenon is alive these days as well. But whether it became better or worse - I don't actually know. It would be a nice thing to look into, although I figure the process of generating the "citation inequality" is so slow (it takes about 10 years or more for a paper to become classics) that we probably just don't have access to old citation networks, as they are probably not fully digitized yet.

Or maybe I'm using wrong keywords.

No comments:

Post a Comment