List<Util.Pair<Integer, Integer>> getCandidates(
Document document, int arg2Line, int connStart, int connEnd, int arg1Line) {
List<Util.Pair<Integer, Integer>> candidates = new ArrayList<Util.Pair<Integer, Integer>>();
int distance = 10;
Sentence arg2Sentence = document.getSentence(arg2Line);
// String conn = arg2Sentence.toString(connStart, connEnd).toLowerCase();
// String category = connAnalyzer.getCategory(conn);
int connHeadPos = connAnalyzer.getHeadWord(arg2Sentence.getParseTree(), connStart, connEnd);
SimpleDepGraph arg2DepGraph = document.getDepGraph(arg2Line);
List<Integer> reachable = arg2DepGraph.getReachableIndices(connHeadPos, false, distance);
for (Integer i : reachable) {
if (arg2Sentence.get(i).getTag("POS").matches("VB.*|NNS?|JJ.*|MD")) {
candidates.add(new Util.Pair<Integer, Integer>(arg2Line, i));
}
}
Tree mainHead = headAnalyzer.getCollinsHead(arg2Sentence.getParseTree().getChild(0));
if (mainHead != null) {
int mainHeadPos = treeAnalyzer.getLeafPosition(arg2Sentence.getParseTree(), mainHead);
List<String> pathAsList = arg2DepGraph.getPathAsList(connHeadPos, mainHeadPos, false);
if (pathAsList != null) {
distance = distance - (1 + pathAsList.size());
} else {
// System.out.println("No path from connHead to mainHead!");
distance--;
}
}
// if (arg1Line == arg2Line) return candidates;
for (int i = arg2Line - 1; i >= 0 && distance >= 0; i--) {
Sentence sentence = document.getSentence(i);
SimpleDepGraph depGraph = document.getDepGraph(i);
mainHead = headAnalyzer.getCollinsHead(sentence.getParseTree().getChild(0));
if (mainHead == null) {
distance--;
continue;
}
int mainHeadPos = treeAnalyzer.getLeafPosition(sentence.getParseTree(), mainHead);
reachable = depGraph.getReachableIndices(mainHeadPos, false, distance);
if (reachable == null) {
distance--;
continue;
}
for (Integer j : reachable) {
if (sentence.get(j).getTag("POS").matches("VB.*|NNS?|JJ.*|MD")) {
candidates.add(new Util.Pair<Integer, Integer>(i, j));
}
}
distance -= 2;
}
return candidates;
}
private PropertyList addDependencyFeatures(
PropertyList pl,
Document doc,
Pair<Integer, Integer> candidate,
int arg2Line,
int arg2HeadPos,
int connStart,
int connEnd) {
Sentence arg2Sentence = doc.getSentence(arg2Line);
String conn = arg2Sentence.toString(connStart, connEnd);
String category = connAnalyzer.getCategory(conn.toLowerCase());
int connHeadPos = connAnalyzer.getHeadWord(arg2Sentence.getParseTree(), connStart, connEnd);
int arg1Line = candidate.first();
Tree arg1Tree = doc.getTree(arg1Line);
int arg1HeadPos = candidate.second();
List<String> path = new ArrayList<String>();
if (arg1Line == arg2Line) {
SimpleDepGraph depGraph = doc.getDepGraph(arg1Line);
List<String> tmpPath = depGraph.getPathAsList(connHeadPos, arg1HeadPos, false);
if (tmpPath != null) {
path.addAll(tmpPath);
} else {
path.add("null");
}
} else {
Tree arg2Root = arg2Sentence.getParseTree();
Tree mainHead = headAnalyzer.getCollinsHead(arg2Root.getChild(0));
int mainHeadPos = treeAnalyzer.getLeafPosition(arg2Root, mainHead);
if (mainHeadPos != -1) {
SimpleDepGraph depGraph = doc.getDepGraph(arg2Line);
List<String> tmpPath = depGraph.getPathAsList(connHeadPos, mainHeadPos, false);
if (tmpPath != null) {
path.addAll(tmpPath);
} else {
path.add("null");
}
}
for (int i = 0; i < Math.abs(arg1Line - arg2Line); i++) {
path.add("SENT");
}
Tree arg1Root = arg1Tree;
mainHead = headAnalyzer.getCollinsHead(arg1Root.getChild(0));
mainHeadPos = treeAnalyzer.getLeafPosition(arg1Root, mainHead);
if (mainHeadPos != -1) {
SimpleDepGraph depGraph = doc.getDepGraph(arg1Line);
List<String> tmpPath = depGraph.getPathAsList(mainHeadPos, arg1HeadPos, false);
if (tmpPath != null) {
path.addAll(tmpPath);
} else {
path.add("null");
}
}
}
StringBuilder sbPath = new StringBuilder();
StringBuilder sbPathWithoutCC = new StringBuilder();
StringBuilder sbPathWithoutRep = new StringBuilder();
String prev = "";
for (String node : path) {
sbPath.append(node).append(":");
if (!node.matches("cc|-cc")) {
sbPathWithoutCC.append(node).append(":");
}
if (!node.equals(prev)) {
sbPathWithoutRep.append(node).append(":");
}
prev = node;
}
// M-dependency path
pl = PropertyList.add("M=" + sbPath.toString(), 1.0, pl);
// Q-M&C
pl = PropertyList.add("Q=" + "CONN-" + conn + '&' + "M-" + sbPath.toString(), 1.0, pl);
// T-M&R
pl = PropertyList.add("T=" + "CAT-" + category + '&' + "M-" + sbPath.toString(), 1.0, pl);
// O-collapsed path without cc
pl = PropertyList.add("O=" + sbPathWithoutCC.toString(), 1.0, pl);
// P-collapsed path without repetition
pl = PropertyList.add("P=" + sbPathWithoutRep.toString(), 1.0, pl);
return pl;
}
private PropertyList addConstituentFeatures(
PropertyList pl,
Document doc,
Pair<Integer, Integer> candidate,
int arg2Line,
int arg2HeadPos,
int connStart,
int connEnd) {
Sentence arg2Sentence = doc.getSentence(arg2Line);
String conn = arg2Sentence.toString(connStart, connEnd);
int connHeadPos = connAnalyzer.getHeadWord(arg2Sentence.getParseTree(), connStart, connEnd);
int arg1Line = candidate.first();
Tree arg1Tree = doc.getTree(arg1Line);
int arg1HeadPos = candidate.second();
List<String> path = new ArrayList<String>();
List<String> pathWithoutPOS = new ArrayList<String>();
if (arg1Line == arg2Line) {
Tree root = arg1Tree;
List<Tree> leaves = root.getLeaves();
List<Tree> treePath = root.pathNodeToNode(leaves.get(connHeadPos), leaves.get(arg1HeadPos));
if (treePath != null) {
for (Tree t : treePath) {
if (!t.isLeaf()) {
path.add(t.value());
if (!t.isPreTerminal()) {
pathWithoutPOS.add(t.value());
}
}
}
}
} else {
Tree arg2Root = arg2Sentence.getParseTree();
Tree mainHead = headAnalyzer.getCollinsHead(arg2Root.getChild(0));
List<Tree> leaves = arg2Root.getLeaves();
int mainHeadPos = treeAnalyzer.getLeafPosition(arg2Root, mainHead);
if (mainHeadPos != -1) {
List<Tree> treePath =
arg2Root.pathNodeToNode(leaves.get(connHeadPos), leaves.get(mainHeadPos));
if (treePath != null) {
for (Tree t : treePath) {
if (!t.isLeaf()) {
path.add(t.value());
if (!t.isPreTerminal()) {
pathWithoutPOS.add(t.value());
}
}
}
}
}
for (int i = 0; i < Math.abs(arg1Line - arg2Line); i++) {
path.add("SENT");
pathWithoutPOS.add("SENT");
}
Tree arg1Root = arg1Tree;
mainHead = headAnalyzer.getCollinsHead(arg1Root.getChild(0));
leaves = arg1Root.getLeaves();
mainHeadPos = treeAnalyzer.getLeafPosition(arg1Root, mainHead);
if (mainHeadPos != -1) {
List<Tree> treePath =
arg1Root.pathNodeToNode(leaves.get(mainHeadPos), leaves.get(arg1HeadPos));
if (treePath != null) {
for (Tree t : treePath) {
if (!t.isLeaf()) {
path.add(t.value());
if (!t.isPreTerminal()) {
pathWithoutPOS.add(t.value());
}
}
}
}
}
}
// H-full path
// L-C&H
StringBuilder fullPath = new StringBuilder();
for (String node : path) {
fullPath.append(node).append(":");
}
pl = PropertyList.add("H=" + fullPath.toString(), 1.0, pl);
pl = PropertyList.add("L=CONN-" + conn + "&" + "H-" + fullPath.toString(), 1.0, pl);
// I-length of path
pl = PropertyList.add("I=" + path.size(), 1.0, pl);
// J-collapsed path without part of speech
// K-collapsed path without repititions
fullPath = new StringBuilder();
StringBuilder collapsedPath = new StringBuilder();
String prev = "";
for (String node : pathWithoutPOS) {
fullPath.append(node).append(":");
if (!node.equals(prev)) {
collapsedPath.append(node).append(":");
}
prev = node;
}
pl = PropertyList.add("J=" + fullPath.toString(), 1.0, pl);
pl = PropertyList.add("K=" + collapsedPath.toString(), 1.0, pl);
return pl;
}