/**
* The basic method for splitting off a clause of a tree. This modifies the tree in place. This
* method addtionally follows ref edges.
*
* @param tree The tree to split a clause from.
* @param toKeep The edge representing the clause to keep.
*/
@SuppressWarnings("unchecked")
private void simpleClause(SemanticGraph tree, SemanticGraphEdge toKeep) {
splitToChildOfEdge(tree, toKeep);
// Follow 'ref' edges
Map<IndexedWord, IndexedWord> refReplaceMap = new HashMap<>();
// (find replacements)
for (IndexedWord vertex : tree.vertexSet()) {
for (SemanticGraphEdge edge : extraEdgesByDependent.get(vertex)) {
if ("ref".equals(edge.getRelation().toString())
&& // it's a ref edge...
!tree.containsVertex(
edge.getGovernor())) { // ...that doesn't already exist in the tree.
refReplaceMap.put(vertex, edge.getGovernor());
}
}
}
// (do replacements)
for (Map.Entry<IndexedWord, IndexedWord> entry : refReplaceMap.entrySet()) {
Iterator<SemanticGraphEdge> iter = tree.incomingEdgeIterator(entry.getKey());
if (!iter.hasNext()) {
continue;
}
SemanticGraphEdge incomingEdge = iter.next();
IndexedWord governor = incomingEdge.getGovernor();
tree.removeVertex(entry.getKey());
addSubtree(
tree,
governor,
incomingEdge.getRelation().toString(),
this.tree,
entry.getValue(),
this.tree.incomingEdgeList(tree.getFirstRoot()));
}
}
private Counter<String> highKLFromMean() {
// Get confidences
Counter<String> kl =
new ClassicCounter<>(MapFactory.<String, MutableDouble>linkedHashMapFactory());
for (Map<SentenceKey, EnsembleStatistics> impl : this.impl) {
for (Map.Entry<SentenceKey, EnsembleStatistics> entry : impl.entrySet()) {
kl.setCount(entry.getKey().sentenceHash, entry.getValue().averageKLFromMean());
}
}
return kl;
}
public SentenceStatistics mean() {
double sumConfidence = 0;
int countWithConfidence = 0;
Counter<String> avePredictions =
new ClassicCounter<>(MapFactory.<String, MutableDouble>linkedHashMapFactory());
// Sum
for (SentenceStatistics stat : this.statisticsForClassifiers) {
for (Double confidence : stat.confidence) {
sumConfidence += confidence;
countWithConfidence += 1;
}
assert Math.abs(stat.relationDistribution.totalCount() - 1.0) < 1e-5;
for (Map.Entry<String, Double> entry : stat.relationDistribution.entrySet()) {
assert entry.getValue() >= 0.0;
assert entry.getValue() == stat.relationDistribution.getCount(entry.getKey());
avePredictions.incrementCount(entry.getKey(), entry.getValue());
assert stat.relationDistribution.getCount(entry.getKey())
== stat.relationDistribution.getCount(entry.getKey());
}
}
// Normalize
double aveConfidence = sumConfidence / ((double) countWithConfidence);
// Return
if (this.statisticsForClassifiers.size() > 1) {
Counters.divideInPlace(avePredictions, (double) this.statisticsForClassifiers.size());
}
if (Math.abs(avePredictions.totalCount() - 1.0) > 1e-5) {
throw new IllegalStateException("Mean relation distribution is not a distribution!");
}
assert this.statisticsForClassifiers.size() > 1
|| this.statisticsForClassifiers.size() == 0
|| Counters.equals(
avePredictions,
statisticsForClassifiers.iterator().next().relationDistribution,
1e-5);
return countWithConfidence > 0
? new SentenceStatistics(avePredictions, aveConfidence)
: new SentenceStatistics(avePredictions);
}
private Counter<String> lowAverageConfidence() {
// Get confidences
Counter<String> lowConfidence =
new ClassicCounter<>(MapFactory.<String, MutableDouble>linkedHashMapFactory());
for (Map<SentenceKey, EnsembleStatistics> impl : this.impl) {
for (Map.Entry<SentenceKey, EnsembleStatistics> entry : impl.entrySet()) {
SentenceStatistics average = entry.getValue().mean();
for (double confidence : average.confidence) {
lowConfidence.setCount(entry.getKey().sentenceHash, 1 - confidence);
}
}
}
return lowConfidence;
}