Arxiv vs Snarxiv(snarxiv.org)
snarxiv.org
Arxiv vs Snarxiv
http://snarxiv.org/vs-arxiv/
7 comments
After falling below 50% trying to reason them out, I went consistently for 'longer title' -- and rose to about 65% correct. Short of actually knowing the fields of study, or searching for the actual titles, there may be no better heuristic.
I tried 10 and got 9 right. I wouldn't claim to "actually know the fields of study", though I am a mathematician so there were a few where I was able to say "no, wait, that wouldn't make any sense" with reasonable confidence. I did see some mostly-syntactic giveaways other than too-short titles (in fact, I think the shorter titles were mostly the genuine ones for me), such as titles that begin "From ..." but don't have any corresponding "to"; titles with unrealistically long chains of "structure" -- "representations of X over Y-modules over Z over W over T" or whatever; overuse of "Towards ...".
They have a context-free grammar, which is translated by a Perl script into OCaml code which "is also fast, allowing the snarXiv to generate papers even more swiftly than your favorite python script, or Ed Witten in the 80’s."
They have a context-free grammar, which is translated by a Perl script into OCaml code which "is also fast, allowing the snarXiv to generate papers even more swiftly than your favorite python script, or Ed Witten in the 80’s."
I went 9 for 11, but the more interesting part is the breakdown on each title. The two I missed fooled 3 out of 3 and 12 out of 16 respectively, and a few of the ones I got right, (nearly) everyone got right.
It seemed to be the ones that included latex math mode were true.
This is wonderful. I am currently "Better than a Monkey."
Just.
Just.
Their 'About' page gives a fun overview of context-free grammars: http://davidsd.org/2010/03/the-snarxiv/
Implementation details and further information: http://davidsd.org/2010/03/the-snarxiv/