/** Load input file */
private void loadInputFile() {
URI_ID = new TObjectIntHashMap<String>();
// load uri--id from input file
TextFileUtils.loadInputURIs(inputFile, URI_ID, false);
logger.debug("Input URIs loading. " + URI_ID.size() + " URIs loaded.");
}
public static TIntObjectHashMap<String> getAllWordIds() {
TObjectIntHashMap<String> wordIds = DataAccess.getAllWordIds();
TIntObjectHashMap<String> idWords = new TIntObjectHashMap<String>(wordIds.size());
for (TObjectIntIterator<String> itr = wordIds.iterator(); itr.hasNext(); ) {
itr.advance();
idWords.put(itr.value(), itr.key());
}
return idWords;
}
public TreeMap<Integer, Integer> addStringFeatureVector(
String[] strFeatures, String label, boolean flagTest) {
HashSet<String> setFeatures = new HashSet<String>();
TreeMap<Integer, Integer> vector = new TreeMap<Integer, Integer>();
for (String feature : strFeatures) {
setFeatures.add(feature);
}
if (setFeatures.size() == 0) return null;
if (!label.equals(""))
if (labels.contains(label)) {
vector.put(labelKey, labels.indexOf(label));
} else {
if (!flagTest) {
labels.add(label);
vector.put(labelKey, labels.indexOf(label));
} else {
// throw new IllegalArgumentException("Label of Testing Data is error!!!");
return null;
}
}
for (String feature : setFeatures) {
if (wordlist.contains(feature)) {
vector.put(wordlist.get(feature), 1);
} else {
if (!flagTest) {
wordlist.put(feature, wordlist.size());
vector.put(wordlist.get(feature), 1);
}
}
}
return vector;
}