Reads collections of text from a set of directories, assigning each directory to a class (as specified by parameter text_directories), and transforms them into a TF-IDF or other word vector. Finally, an SVM is applied to model the input texts.