// generating features
public void generateFeatures() {
if (features != null) {
features.clear();
} else {
features = new ArrayList();
}
if (fmap != null) {
fmap.clear();
} else {
fmap = new HashMap();
}
if (currentFeatures != null) {
currentFeatures.clear();
} else {
currentFeatures = new ArrayList();
}
if (data.trnData == null || dict.dict == null) {
System.out.println("No data or dictionary for generating features");
return;
}
// scan over data list
for (int i = 0; i < data.trnData.size(); i++) {
Observation obsr = (Observation) data.trnData.get(i);
for (int j = 0; j < obsr.cps.length; j++) {
Element elem;
CountFIdx cntFIdx;
elem = (Element) dict.dict.get(new Integer(obsr.cps[j]));
if (elem != null) {
if (elem.count <= option.cpRareThreshold) {
// skip this context predicate, it is too rare
continue;
}
cntFIdx = (CountFIdx) elem.lbCntFidxes.get(new Integer(obsr.humanLabel));
if (cntFIdx != null) {
if (cntFIdx.count <= option.fRareThreshold) {
// skip this feature, it is too rare
continue;
}
} else {
// not found in the dictionary, then skip
continue;
}
} else {
// not found in the dictionary, then skip
continue;
}
// update the feature
Feature f = new Feature(obsr.humanLabel, obsr.cps[j]);
f.strId2Idx(fmap);
if (f.idx < 0) {
// new feature, add to the feature list
addFeature(f);
// update the feature index in the dictionary
cntFIdx.fidx = f.idx;
elem.chosen = 1;
}
}
}
option.numFeatures = features.size();
}
