@Override
public SparseFeatureVector calculateFeature(
Instance instance,
Label y,
int lastIndex,
int target,
LabelUnit candidateLabelUnit,
boolean local) {
// TODO cache strings
SparseFeatureVector fv = new SparseFeatureVector(params);
if (instance.getSequence().get(target) instanceof Word) {
Word word = (Word) instance.getSequence().get(target);
String candidateLabel = ((WordLabelUnit) candidateLabelUnit).getLabel();
assert word.getWord().getCache() != null
: word.getWord().getId() + ":" + word.getWord().getRealBase();
fv.add(word.getWord().getCache().getNodeFeatures(), candidateLabel);
if (params.getUseGlobalFeatures() && !local) {
SparseFeatureVector globalFv =
calcWordGlobalFeatures(
word, instance, y, lastIndex, (WordLabelUnit) candidateLabelUnit);
assert params.getAverageNumWords() > 0.;
globalFv.scale(params.getGlobalWeight());
fv.add(globalFv, "GLOBAL");
}
} else {
assert instance.getSequence().get(target) instanceof Pair;
Pair pair = (Pair) instance.getSequence().get(target);
PairLabelUnit candidatePairLabelUnit = (PairLabelUnit) candidateLabelUnit;
Node w1 = pair.getW1().getWord();
Node w2 = pair.getW2().getWord();
fv.add(
w1.getCache().getFullPathFeatures(w2), candidatePairLabelUnit.getLabel().concat("PATH"));
if (params.getUseGlobalFeatures() && !local) {
SparseFeatureVector globalFv =
calcPairGlobalFeatures(pair, instance, y, lastIndex, candidatePairLabelUnit);
assert params.getAverageNumWords() > 0.;
globalFv.scale(params.getGlobalWeight());
fv.add(globalFv, "GLOBAL");
}
}
assert fv.size() > 0;
return fv;
}
protected void addParallelInfoToFV(
SparseFeatureVector fv,
String wBase,
String wPOS,
Node w1,
String label1,
Node w2,
String label2,
String header) {
SparseFeatureVector paraFv = new SparseFeatureVector(params);
paraFv.add(
w1.getCache().getShortestPathFeatures(w2),
header.concat(label1).concat(wBase).concat(label2));
paraFv.add(
w1.getCache().getShortestPathFeatures(w2),
header.concat(label1).concat(wPOS).concat(label2));
paraFv.add(w1.getCache().getShortestPathFeatures(w2), header.concat(label1).concat(label2));
paraFv.normalize(1.);
fv.add(paraFv);
}