private void createTestFile(String output, Vector bestWordsForm, StringBuffer header)
throws FileNotFoundException, IOException {
OutputStream fout = new FileOutputStream(output + "_test", false);
OutputStream bout = new BufferedOutputStream(fout);
OutputStreamWriter outputFile = new OutputStreamWriter(bout);
StringBuffer tail = new StringBuffer();
tail.append("\n");
tail.append("\n");
tail.append("@DATA");
tail.append("\n");
// System.out.println("Print positve Test...");
for (int i = 0; i < positiveTestExamples.length; i++) {
VSMVector formTemp = positiveTestExamples[i];
// formTemp.squaredNormalization();
tail.append("{");
for (int j = 0; j < attributes.size(); j++) {
VSMElement elemForm = formTemp.getElement(attributes.elementAt(j));
if (elemForm != null) {
tail.append(j);
tail.append(" ");
tail.append((int) elemForm.getWeight());
tail.append(",");
}
}
tail.append(attributes.size() + " S}");
tail.append("\n");
}
// System.out.println("Print negative Test...");
for (int i = 0; i < negativeTestExamples.length; i++) {
VSMVector formTemp = negativeTestExamples[i];
// formTemp.squaredNormalization();
tail.append("{");
for (int j = 0; j < attributes.size(); j++) {
VSMElement elemForm = formTemp.getElement(attributes.elementAt(j));
if (elemForm != null) {
tail.append(j);
tail.append(" ");
tail.append((int) elemForm.getWeight());
tail.append(",");
}
}
tail.append(attributes.size() + " NS}");
tail.append("\n");
}
outputFile.write(header.toString());
outputFile.flush();
outputFile.write(tail.toString());
outputFile.close();
}
protected VSMVector[] createVSM(
File[] files, StopList stoplist, int[] indexes, boolean addToFeatures) throws SAXException {
Vector<VSMVector> tempVSM = new Vector<VSMVector>();
for (int i = 0; i < files.length && i < indexes.length; i++) {
try {
VSMVector vsm = new VSMVector(files[indexes[i]].toString(), isForm, stoplist);
tempVSM.add(vsm);
if (addToFeatures) {
Iterator iterator1 = vsm.getElements();
while (iterator1.hasNext()) {
VSMElement elem = (VSMElement) iterator1.next();
VSMElement value = (VSMElement) df.get(elem.getWord());
if (value == null) {
df.put(elem.getWord(), new VSMElement(elem.getWord(), 1));
} else {
df.put(elem.getWord(), new VSMElement(elem.getWord(), value.getWeight() + 1));
}
}
}
// System.out.println(vsm.toString());
} catch (IOException ex) {
}
}
VSMVector[] examples = new VSMVector[tempVSM.size()];
tempVSM.toArray(examples);
return examples;
}
protected VSMVector[] createVSM(File file, StopList stoplist) throws SAXException {
Vector<VSMVector> tempVSM = new Vector<VSMVector>();
try {
BufferedReader reader = new BufferedReader(new InputStreamReader(new FileInputStream(file)));
for (String line = reader.readLine(); line != null; line = reader.readLine()) {
VSMVector vsm = new VSMVector(line, stoplist);
tempVSM.add(vsm);
Iterator iterator1 = vsm.getElements();
while (iterator1.hasNext()) {
VSMElement elem = (VSMElement) iterator1.next();
VSMElement value = (VSMElement) df.get(elem.getWord());
if (value == null) {
df.put(elem.getWord(), new VSMElement(elem.getWord(), 1));
} else {
df.put(elem.getWord(), new VSMElement(elem.getWord(), value.getWeight() + 1));
}
}
// System.out.println(vsm.toString());
}
} catch (IOException ex) {
ex.printStackTrace();
}
VSMVector[] examples = new VSMVector[tempVSM.size()];
tempVSM.toArray(examples);
return examples;
}
public String[] centroid2Weka(String output) throws FileNotFoundException, IOException {
// FileOutputStream fout = new FileOutputStream(output,false);
// DataOutputStream dout = new DataOutputStream( fout );
OutputStream fout = new FileOutputStream(output, false);
OutputStream bout = new BufferedOutputStream(fout);
OutputStreamWriter outputFile = new OutputStreamWriter(bout);
StringBuffer header = new StringBuffer();
header.append("@RELATION TSFC");
header.append("\n");
header.append("\n");
StringBuffer tail = new StringBuffer();
// Iterator iterator1 = centroidForm.getElements();
// Vector bestWordsForm = new Vector();
// while (iterator1.hasNext()) {
// bestWordsForm.add(iterator1.next());
// }
Vector bestWordsForm = new Vector(df.values());
Collections.sort(bestWordsForm, new VSMElementComparator());
// for (int i = 0; i < 10; i++) {
// VSMElement elem1 = (VSMElement)bestWordsForm.elementAt(i);
// System.out.println(elem1.getWord());
// System.out.println(elem1.getWeight());
// }
// System.out.println("Print header...");
for (int i = 0; i <= numOfFeatures && i < bestWordsForm.size(); i++) {
VSMElement elem = (VSMElement) bestWordsForm.elementAt(i);
if (elem.getWeight() > minDF) {
header.append("@ATTRIBUTE ");
header.append(elem.getWord());
attributes.add(elem.getWord());
header.append(" REAL");
header.append("\n");
}
}
header.append("@ATTRIBUTE class {S,NS}");
tail.append("\n");
tail.append("\n");
tail.append("@DATA");
tail.append("\n");
// System.out.println("Print positive...");
for (int i = 0; i < positiveExamples.length; i++) {
VSMVector formTemp = positiveExamples[i];
// formTemp.squaredNormalization();
tail.append("{");
for (int j = 0; j < attributes.size(); j++) {
VSMElement elemForm = formTemp.getElement(attributes.elementAt(j));
if (elemForm != null) {
tail.append(j);
tail.append(" ");
tail.append((int) elemForm.getWeight());
tail.append(",");
}
}
tail.append(attributes.size() + " S}");
tail.append("\n");
}
// System.out.println("Print negative...");
for (int i = 0; i < negativeExamples.length; i++) {
VSMVector formTemp = negativeExamples[i];
// formTemp.squaredNormalization();
tail.append("{");
for (int j = 0; j < attributes.size(); j++) {
VSMElement elemForm = formTemp.getElement(attributes.elementAt(j));
if (elemForm != null) {
tail.append(j);
tail.append(" ");
tail.append((int) elemForm.getWeight());
tail.append(",");
}
}
tail.append(attributes.size() + " NS}");
tail.append("\n");
}
outputFile.write(header.toString());
outputFile.flush();
outputFile.write(tail.toString());
outputFile.close();
if (positiveTestExamples != null) {
createTestFile(output, bestWordsForm, header);
}
String[] atts = new String[attributes.size()];
attributes.toArray(atts);
return atts;
}