As I wrote in the Zotero plugins forum, I don't seem to be able to get this to work. Could this be another one of the remaining 'path issues' that you mention here? And if so, is there anything we can do to help in debugging it? I am currently running a process with the debugging option on, and will upload the output to gits in a while.
The debugging output will probably not contain the error, as (from what you describe on the Zotero forums) the problem is with the Python script. These scripts keep logs in the zotero\papermachines\out\logs directory in your Firefox profile folder.
It would obviously make a lot more sense to have this output displayed directly when a process fails rather than asking people to root around for it, so I'll implement that for the next version. I'll also add an interface to adjust the parameters of the process and save preferences for it.
In the meantime, would you mind looking in the mallet_lda.log file and seeing what errors are present there? It might be quite large, since the program automatically runs again if it detects that the process has failed (hence, the hours for which it appeared to run might have actually been the process continually restarting.)
I uploaded the log to https://gist.github.com/3778779 . Does that work?
Thanks, I see it now. That line "INFO:root:LDA complete in 0.069000005722 seconds" leads me to suspect that Mallet isn't actually running. I'll check into it on my virtual Windows machine and see what I find.
Great. Thanks much.
Indeed, Java on Windows was failing both due to the superfluous "-server" option and the fact that classpaths are parsed differently on the Windows command line.
I'll quickly put together the interface to adjust some of the model parameters and upload a fixed version.
Fantastic! Can't wait to start playing with it... And if I find some time this weekend, I WILL make a little camtasia video with it, just to show people how it works and what it can be used for,
Oh and by the bye Chris, could you possibly also doublecheck wordclouds (https://gist.github.com/3779424 )and classifier (https://gist.github.com/3779419 ) for possible similar mistakes (under windows)? Because those also choke with me - only geoparser and dbpedia work perfeccty for me...
Thanks for your patience! Classifier issue was due to the same Java problem. Word clouds were failing when there were no words found for a particular time interval/subcollection, either due to filtering or sparsity of data. These should be fixed in the latest version, 0.2.1.