@Override
public void annotate(Blackboard blackboard, DocumentComponent component) {
// get the grams and order them by score
Collection<Gram> grams = blackboard.getKeyphrases();
Map<Keyphrase, Double> scoredGrams = new HashMap<>();
for (Gram g : grams) {
Keyphrase k = (Keyphrase) g;
scoredGrams.put(k, k.getFeature(GenericEvaluatorAnnotator.SCORE));
}
List<Map.Entry<Keyphrase, Double>> gramsToTrash =
scoredGrams
.entrySet()
.stream()
.sorted(Collections.reverseOrder(Map.Entry.comparingByValue()))
.collect(Collectors.toList());
// now the have the gram ordered by score.
// we want to keep at least the first 5%; then, we look for the maximum
// steep in score in the following 20%, and we discard everything after
// that remains.
double bestRange = Math.ceil((gramsToTrash.size() * 5.0) / 100.0);
double steepRange = Math.ceil((gramsToTrash.size() * 25.0) / 100.0);
// keep the first 5%
for (int i = 0; i < bestRange; i++) {
gramsToTrash.remove(0);
}
double maxSteep = Double.MIN_VALUE;
int maxSteepIndex = Integer.MIN_VALUE;
// search for the maximum steep in the next 15%
for (int i = 0; i < steepRange - 1; i++) {
double steep = (gramsToTrash.get(i).getValue() - gramsToTrash.get(i + 1).getValue());
if (steep > maxSteep) {
maxSteep = steep;
maxSteepIndex = i;
}
}
// keep the grams before the steep
for (int i = 0; i < maxSteepIndex; i++) {
gramsToTrash.remove(0);
}
// now remove the remaining grams from the blackboard.
for (Map.Entry<Keyphrase, Double> e : gramsToTrash) blackboard.removeKeyphrase(e.getKey());
}
@Override
default void run(Blackboard b) {
this.annotate(b, b.getStructure());
}
