Use of probabilistic graphical Models to predict forum posts in avsforum.com. Project for CS5340

##Directory structure

1. working: Stores working files for use later. e.g.

   • pickled models to be used for evaluation

2. obj_globals: global objects to be used for accessing data quickly

   • pickled LDA model. Topics, tokens etc.

3. reports: Generated files showing performance of models

   • TSVs
   • CSVs
   • partial .tex files of tables

##Preprocessing

1. learn_topics topics learnt from data.
2. topic_time_dist generates arrays of time histograms of topics. We're looking for time distributions significantly different between topics.