/**
* TODO move this to Utils.java because other classes have a similar method Creates a file
* containing the documents to be used for training Documents are represented as an n x m matrix,
* where each row is normalized (ie each entry i, j is the percentage of document i that is word
* j)
*
* @param filename The name of the file to be created
* @param papers The list of documents to be used for training
*/
private void createKmeansInput(String filename, List<TrainingPaper> papers) {
System.out.print("creating kmeans input: " + filename + " ... ");
double[][] fullRepresentation = new double[papers.size()][terms.size()];
int paperIndex = 0;
for (TrainingPaper paper : papers) {
double sum = 0;
for (int word : paper.getTrainingWords()) {
double wordFrequency = paper.getTrainingTf(word);
fullRepresentation[paperIndex][word] = wordFrequency;
sum += wordFrequency;
}
for (int col = 0; col < fullRepresentation[paperIndex].length; col++) {
fullRepresentation[paperIndex][col] /= sum;
}
paperIndex++;
}
PlusoneFileWriter fileWriter = new PlusoneFileWriter(filename);
for (int row = 0; row < fullRepresentation.length; row++) {
for (int col = 0; col < fullRepresentation[row].length; col++) {
fileWriter.write(fullRepresentation[row][col] + " ");
}
fileWriter.write("\n");
}
fileWriter.close();
System.out.println("done.");
}
/**
* TODO This method exists in several classes--move to Utils? Takes a list of PaperAbstract
* documents and writes them to file according to the format specified by lda-c-dist
*
* @param filename name of the file to be created (will be overwritten if it already exists)
* @param papers list of papers to be written to file
*/
private void createLdaInputTest(String filename, List<PredictionPaper> papers) {
System.out.print("creating lda test input in file: " + filename + " ... ");
PlusoneFileWriter fileWriter = new PlusoneFileWriter(filename);
for (PredictionPaper paper : papers) {
fileWriter.write(paper.getTrainingWords().size() + " ");
for (int word : paper.getTrainingWords()) {
fileWriter.write(word + ":" + paper.getTrainingTf(word) + " ");
}
fileWriter.write("\n");
}
fileWriter.close();
System.out.println("done.");
}