[PRL] Is POPL really one research community?
Mitchell Wand
wand at ccs.neu.edu
Tue Oct 21 20:37:22 EDT 2014
http://www.pl-enthusiast.net/2014/10/21/is-popl-one-community/
*Methodology*
The concrete questions that I wanted to investigate were:
- How closely connected is the POPL community? As in, suppose you
construct a*collaboration graph* where nodes represent POPL authors, and
edges link authors who have co-authored a POPL paper. Or alternatively,
suppose you built an *overlap graph *where nodes are POPL papers with
multiple authors, and edges connect papers that share one or more
authors. What do these graphs look like?
- Specifically, are the graphs made of a few disjoint components?
- Are there prominent clusters of authors who publish POPL papers
together, or clusters of papers that are written by a subcommunity of
researchers?
Note that the collaboration graph and the overlap graph are closely
related. A node in the overlap graph — a paper — corresponds to a
*clique,* consisting
of the paper’s authors, in the collaboration graph. This means that
the overlap graph is an *abstraction *of the collaboration graph, and you
can easily construct the latter from the former. Empirically, if you take
the collaboration graph and coalesce maximal cliques in it into single
nodes, you get the overlap graph. However, the overlap graph is easier to
visualize because it has fewer nodes and cliques, and because of this, I
will mostly focus on this graph in the rest of this post.
-------------- next part --------------
HTML attachment scrubbed and removed
More information about the PRL
mailing list