Edges represent the tropes that movies have in common. Similar
movies are closer together.
This page may not work in some versions of IE. It's fine in at
least firefox and chrome.
Data comes from
tvtropes.org,
which is parsed nightly by the
DBTropes project. I
picked up that result, loaded it into a local 4store database.
I count all the shared tropes in many movies (but not all of them yet--
that will need to be done in a more efficient way).
Here's the code and the SPARQL queries. At first I
wrote the output in a way that gephi could read, but I am not good enough at gephi to do anything more than follow these instructions.
The page you're looking at reads the JSON output of the
counter program and uses the
D3 javascript library
to help with the drawing. D3 comes with a force layout algorithm
(cool demo),
but I had trouble adapting it to do per-edge lengths,
so I adapted the algorithm from
jsvggraph
instead. Arbor looks cool too, but I didn't try it.
What this tool mostly needs now is a way to control which
movies you're looking at (all 2010 movies, all Bruce Willis
movies, movies featuring trope T, etc). But even with the
semi-random set (pulled from a "top 500 movies" list), you can
still make pretty pictures like this.
See also
Recent changes to this project
Intro to tvtropes for semantic web practitioners (or anyone else)
Discussion of tropeGraph on the tvtropes forum
