2011-10-01  Timothy P. Jurka  <tpjurka@ucdavis.edu>

	* DESCRIPTION: Release 1.3.2
	* Added CITATION
	* Fixed bugs with maxent function in cross_validate()
	* Fixed memory leak when dealing with factors instead of vectors
	* Updated train_models() function to support new maxent implementation
	* Updated DESCRIPTION file

2011-08-23  Timothy P. Jurka  <tpjurka@ucdavis.edu>

	* DESCRIPTION: Release 1.3.1
	* Added ChangeLog
	* Added inst/
	* Added inst/examples
	* Removed RTextTools-internal.R
	* Removed accuracy.R
	* Removed confusion_create.R
	* Removed dtm_to_sparsem.R (see as.compressed.matrix in package maxent)
	* Removed read_access.R
	* Removed score_classify.R
	* Updated LogitBoost train function to prevent certain bugs in Linux

2011-08-22  Timothy P. Jurka  <tpjurka@ucdavis.edu>

	* DESCRIPTION: Release 1.3
	* Added create_scoreSummary()
	* Added create_scoreSummary() documentation
	* Added Rstem as a CRAN package
	* Removed score_classify()
	* Removed references to Naive Bayes in the documentation
	* Updated create_precisionRecallSummary.R with create_scoreSummary()
	* Updated create_analytics.R with create_scoreSummary()
	* Updated stemming function to fix poor stemming performance
        
2011-08-16  Timothy P. Jurka  <tpjurka@ucdavis.edu>

	* DESCRIPTION: Release 1.2
	* Passes R CMD CHECK, meaning it can be submitted to CRAN
	* Added links to functions in documentation for easier navigation, expanded documentation to cover all functions and datasets
	* Added URL to DOCUMENTATION file, standardized description with what's on the website
	* Removed extraneous code (unused functions, commented code, etc.)
	* Removed dtm_to_sparsem and use the as.compressed.matrix() function in maxent instead
	* Changed glmnet iteration parameter to 10^5
	* Changed classify_model() ... parameter to be passed into all predict() functions
        
2011-08-02  Timothy P. Jurka  <tpjurka@ucdavis.edu>

	* DESCRIPTION: Release 1.1
	* GLMNET algorithm now uses compressed sparse matrices
	* Full user documentation added
	* User interface simplified, less parameters need to be passed in
	* Bundled two sample datasets: NYTimes and USCongress
	* Better analytics, with precision, recall, and F1 scores
	* Simpler installation / Removed rJava
        
2011-06-18  Timothy P. Jurka  <tpjurka@ucdavis.edu>

	* DESCRIPTION: Release 1.0
        