PhiloMine

PhiloMine (release 0.1a)
Future Plans
ARTFL/DLDC
University of Chicago

Current work planned includes (but is not limited to):

development of Latent Semantic Index and/or Probabilistic LSI vector space searching;
adding some predictive functions to the all document flavor of PhiloMine, which will probably be done when we have a data set that would benefit from this;
adding the stemmer to the base version of PhiloMine and possibly a simple English stemmer;
saving processed ARFF (Weka) files and possibly processed vector space matricies, rather than recalculating them on each run (code is mainly done);
save SVMLite and Weka SMO models for local, non-WWW processing, an option we may need to do larger runs on more powerful machines once we have tested a feature set, parameters, etc. interactively.

Longer term notions include expanding available kinds of feature sets, such as bi/tri grams, lemmas and port of speech. As currently implemented, these function would simply be stored as alternative base data in the same format as we have surface forms and should be plug compatible.

As we continue working on our own research projects, we expect that other notions will occur to us. Let us know if you have any ideas, and even better, if you have any code.