polo | Polo : A way to play with Mallet

To get started using Polo, the first thing to do is grab the source code from github and clone it into a local directory. You can call the directory anything; polo_root is a good choice. Change directories into the cloned code base and edit the file config-play.ini in your favorite editor:.
There is a project directory where all of your source data and generated output will live. By default, this directory lives in your Polo application root and is called projects. It is automatically created for you when you install Polo. The project directory contains three resources:
A corpus directory, which contains your corpus files and extra stopwords list. By convention, your corpus file is called corpus.csv and is a comma delimmited file with three columns -- a unique document ID, a label of some kind (which must be there, even if it is something you have to make up), and the 'document' itself, which for a topic model is just the unit of text you are analyzing, which may be a paragraph or any other text segment, and not necessarily a stand-alone document. The extra stopwords file is called extra-stopwords.txt and contains stopwords beyond those used by MALLET itself. Note that the stopwords file must exist, even if it is empty, and it must be named as listed here. Ditto for the corpus file (although it better have some content, right?)
A trials directory, which contains subdirectories for each of your topic model trials. Each time you want to run a trial, you create a subdirectory -- say trial1 -- and then put an entry for that trial in the project's config.ini file (see next item), The trials directory is where Polo will put your resulting SQLite database.
A config.ini file to define some things about your project and specific parameters for each. Users of MALLET will recognize that the keys in the trials section are just the command line keys for MALLET's train topics function. Given this, you can add more keys if you'd like. Note, however, that Polo takes care of defining the output files, so you don't need to add these.