private void writeFile(SpectralClustering cluster) {
FileUtil.clean(output);
int[] lab = cluster.getClusterLabel();
for (int x = 0; x < lab.length; x++) {
final Doc doc = list.get(x);
String one = doc.url + "\n" + doc.title + "\n" + doc.text + "\n" + hr;
FileUtil.write(new File(output, lab[x] + ".txt"), one, true);
}
}
public static void clean(File output) {
if (output.isDirectory()) {
for (File f : output.listFiles()) {
clean(f);
f.delete();
}
} else {
output.delete();
}
}
private void _dimension() {
FileUtil.clean(root);
int x = 0;
for (File file : root1.listFiles()) {
Doc doc = buildDoc(x, file);
// System.out.println(doc.vector.getTFIDFValues());
Iterator<Map.Entry<String, Double>> it = doc.vector.getTFIDFValues().entrySet().iterator();
for (; it.hasNext(); ) {
if (it.next().getValue() < 0.05) {
it.remove();
}
}
String str = StringUtils.join(doc.vector.getTFIDFValues().keySet().iterator(), " ");
str = doc.url + "\n" + doc.title + "\n" + str + "\n" + doc.text;
FileUtil.write(new File(root, file.getName()), str, false);
x++;
}
}