JSTORr (version 1.0.20161214)

JSTOR_MALLET: Generate one or more topic models using MALLET's implementation of Latent Dirichlet allocation (LDA)

Description

Generates one or more topic models using MALLET and plots diagnostics. This is a very basic R wrapper for MALLET on Windows, see here for something more complete: http://cran.r-project.org/web/packages/mallet/ For use with JSTOR's Data for Research datasets (http://dfr.jstor.org/).

Usage

JSTOR_MALLET(nouns, MALLET = "C:/mallet-2.0.7", JAVA = "C:/Program Files (x86)/Java/jre7/bin", K)

Arguments

nouns

the object returned by the function JSTOR_dtmofnouns. If you want to use the document contents with all other parts of speech, use unpack1grams$wordcounts as the first argument here (where unpack1grams is the output from JSTOR_unpack1grams)

MALLET

the directory containing MALLET's bin directory, ideally "C:/mallet-2.0.7" or similarly close to C:/ on a Windows computer.

JAVA

the directory containing java.exe. It will probably be something like "C:/Program Files/Java/jre7/bin" but you should check

K

the number of topics that the model should contain. Can also be a vector of numbers of topics, then a model will be generated for each number. Useful for comparing diagnostics of different models, but may be time consuming.

Value

Returns a folder of text files (one text file per document) and a folder of output files from MALLET. The folders are named with a date-time stamp so they wont be overwritten by repeated runs.

Examples

## JSTOR_MALLET(nouns, MALLET = "C:/mallet-2.0.7", JAVA = "C:/Program Files (x86)/Java/jre7/bin", K = 150) # generate a single model
## JSTOR_MALLET(nouns =  unpack1grams$wordcounts, MALLET = "C:/mallet-2.0.7", JAVA = "C:/Program Files (x86)/Java/jre7/bin", K = seq(150, 500, 50)) # can also generate multiple models with different numbers of topics