Travis Hoppe & Dan Appella
Motivation:
Author count histogram for a single issue of JBC
Purpose of this talk is to examine the trends of
journal article authorship
and other interesting facts from the front-end of a journal.
JCI (J. Clinical Investigations), PNAS (Proc. National Academy of Sciences), JACS (J. American Chemical Society), JBC (J .Biological Chemistry), Science, Nature, Cell.
For each journal, every table of contents was downloaded,
this gave title and authorship information for each paper.
Filtered for editorials, corrections, errata, book reviews, ...
Approximately 2.4K TOC and 655K papers.
Why not examine the abstracts, or the articles themselves?
LARGE project collaborations, removed from study.
Finishing the euchromatic sequence of the human genome
Characterization of the Martian Surface Deposits by the Mars Pathfinder Rover, Sojourner
In most cases the distribution of authorship follows a Poisson;
the mean is approximately the variance.
Common metrics for scientific impact:
Total citations
Impact factor / Article influence
Eigenfactor (similar to Google's page-rank)
a priori not the same trend as mean authorship...
Look for significant keywords in the titles
e.g. [the,first,has]
(acids, acidic) acid
Identify a canonical representative set for related words
spectrum 4.24
structure 3.15
neutron 3.01
radioactivity 2.74
NATURE 1951
acid 7.32
chromatography 2.76
structure 2.68
cell 2.20
NATURE 2013
cell 12.59
structure 5.87
protein 5.13
human 5.01
Some metrics: science budgets, big data, nationalities of postdoctoral students, number of international collaborations, ...