/**
* A helper to add an entire subtree to a given dependency tree.
*
* @param toModify The tree to add the subtree to.
* @param root The root of the tree where we should be adding the subtree.
* @param rel The relation to add the subtree with.
* @param originalTree The orignal tree (i.e., {@link ClauseSplitterSearchProblem#tree}).
* @param subject The root of the clause to add.
* @param ignoredEdges The edges to ignore adding when adding this subtree.
*/
private static void addSubtree(
SemanticGraph toModify,
IndexedWord root,
String rel,
SemanticGraph originalTree,
IndexedWord subject,
Collection<SemanticGraphEdge> ignoredEdges) {
if (toModify.containsVertex(subject)) {
return; // This subtree already exists.
}
Queue<IndexedWord> fringe = new LinkedList<>();
Collection<IndexedWord> wordsToAdd = new ArrayList<>();
Collection<SemanticGraphEdge> edgesToAdd = new ArrayList<>();
// Search for subtree to add
for (SemanticGraphEdge edge : originalTree.outgoingEdgeIterable(subject)) {
if (!ignoredEdges.contains(edge)) {
if (toModify.containsVertex(edge.getDependent())) {
// Case: we're adding a subtree that's not disjoint from toModify. This is bad news.
return;
}
edgesToAdd.add(edge);
fringe.add(edge.getDependent());
}
}
while (!fringe.isEmpty()) {
IndexedWord node = fringe.poll();
wordsToAdd.add(node);
for (SemanticGraphEdge edge : originalTree.outgoingEdgeIterable(node)) {
if (!ignoredEdges.contains(edge)) {
if (toModify.containsVertex(edge.getDependent())) {
// Case: we're adding a subtree that's not disjoint from toModify. This is bad news.
return;
}
edgesToAdd.add(edge);
fringe.add(edge.getDependent());
}
}
}
// Add subtree
// (add subject)
toModify.addVertex(subject);
toModify.addEdge(
root,
subject,
GrammaticalRelation.valueOf(Language.English, rel),
Double.NEGATIVE_INFINITY,
false);
// (add nodes)
wordsToAdd.forEach(toModify::addVertex);
// (add edges)
for (SemanticGraphEdge edge : edgesToAdd) {
assert !toModify.incomingEdgeIterator(edge.getDependent()).hasNext();
toModify.addEdge(
edge.getGovernor(),
edge.getDependent(),
edge.getRelation(),
edge.getWeight(),
edge.isExtra());
}
}
/**
* Fix some bizarre peculiarities with certain trees. So far, these include:
*
* <ul>
* <li>Sometimes there's a node from a word to itself. This seems wrong.
* </ul>
*
* @param tree The tree to clean (in place!).
* @return A list of extra edges, which are valid but were removed.
*/
public static List<SemanticGraphEdge> cleanTree(SemanticGraph tree) {
// assert !isCyclic(tree);
// Clean nodes
List<IndexedWord> toDelete = new ArrayList<>();
for (IndexedWord vertex : tree.vertexSet()) {
// Clean punctuation
if (vertex.tag() == null) {
continue;
}
char tag = vertex.backingLabel().tag().charAt(0);
if (tag == '.' || tag == ',' || tag == '(' || tag == ')' || tag == ':') {
if (!tree.outgoingEdgeIterator(vertex)
.hasNext()) { // This should really never happen, but it does.
toDelete.add(vertex);
}
}
}
toDelete.forEach(tree::removeVertex);
// Clean edges
Iterator<SemanticGraphEdge> iter = tree.edgeIterable().iterator();
while (iter.hasNext()) {
SemanticGraphEdge edge = iter.next();
if (edge.getDependent().index() == edge.getGovernor().index()) {
// Clean self-edges
iter.remove();
} else if (edge.getRelation().toString().equals("punct")) {
// Clean punctuation (again)
if (!tree.outgoingEdgeIterator(edge.getDependent())
.hasNext()) { // This should really never happen, but it does.
iter.remove();
}
}
}
// Remove extra edges
List<SemanticGraphEdge> extraEdges = new ArrayList<>();
for (SemanticGraphEdge edge : tree.edgeIterable()) {
if (edge.isExtra()) {
if (tree.incomingEdgeList(edge.getDependent()).size() > 1) {
extraEdges.add(edge);
}
}
}
extraEdges.forEach(tree::removeEdge);
// Add apposition edges (simple coref)
for (SemanticGraphEdge extraEdge :
new ArrayList<>(extraEdges)) { // note[gabor] prevent concurrent modification exception
for (SemanticGraphEdge candidateAppos : tree.incomingEdgeIterable(extraEdge.getDependent())) {
if (candidateAppos.getRelation().toString().equals("appos")) {
extraEdges.add(
new SemanticGraphEdge(
extraEdge.getGovernor(),
candidateAppos.getGovernor(),
extraEdge.getRelation(),
extraEdge.getWeight(),
extraEdge.isExtra()));
}
}
for (SemanticGraphEdge candidateAppos : tree.outgoingEdgeIterable(extraEdge.getDependent())) {
if (candidateAppos.getRelation().toString().equals("appos")) {
extraEdges.add(
new SemanticGraphEdge(
extraEdge.getGovernor(),
candidateAppos.getDependent(),
extraEdge.getRelation(),
extraEdge.getWeight(),
extraEdge.isExtra()));
}
}
}
// Brute force ensure tree
// Remove incoming edges from roots
List<SemanticGraphEdge> rootIncomingEdges = new ArrayList<>();
for (IndexedWord root : tree.getRoots()) {
for (SemanticGraphEdge incomingEdge : tree.incomingEdgeIterable(root)) {
rootIncomingEdges.add(incomingEdge);
}
}
rootIncomingEdges.forEach(tree::removeEdge);
// Loop until it becomes a tree.
boolean changed = true;
while (changed) { // I just want trees to be trees; is that so much to ask!?
changed = false;
List<IndexedWord> danglingNodes = new ArrayList<>();
List<SemanticGraphEdge> invalidEdges = new ArrayList<>();
for (IndexedWord vertex : tree.vertexSet()) {
// Collect statistics
Iterator<SemanticGraphEdge> incomingIter = tree.incomingEdgeIterator(vertex);
boolean hasIncoming = incomingIter.hasNext();
boolean hasMultipleIncoming = false;
if (hasIncoming) {
incomingIter.next();
hasMultipleIncoming = incomingIter.hasNext();
}
// Register actions
if (!hasIncoming && !tree.getRoots().contains(vertex)) {
danglingNodes.add(vertex);
} else {
if (hasMultipleIncoming) {
for (SemanticGraphEdge edge : new IterableIterator<>(incomingIter)) {
invalidEdges.add(edge);
}
}
}
}
// Perform actions
for (IndexedWord vertex : danglingNodes) {
tree.removeVertex(vertex);
changed = true;
}
for (SemanticGraphEdge edge : invalidEdges) {
tree.removeEdge(edge);
changed = true;
}
}
// Return
assert isTree(tree);
return extraEdges;
}