private void getThirdThirdSupArgFeats(
FeaturizedToken pred,
FeaturizedToken arg,
FeaturizedToken predLast,
FeaturizedToken predNext,
FeaturizedToken argLast,
FeaturizedToken argNext,
FeaturizedToken argParent,
ArrayList<FeaturizedToken> argChildrenObjectList,
ArrayList<String> feats) {
String feat;
// a.lowSupportVerb.lemma
feats.add(getFeatureObject(arg.getLowSupportNoun()).getLemma());
// a.lemma + a.h.form
feats.add(arg.getLemma() + "_" + argParent.getForm());
// a.lemma + a.pphead.form
feats.add(arg.getLemma() + "_" + getFeatureObject(arg.getParent()).getForm());
// a1.lemma
feats.add(argNext.getLemma());
// a.pos + a.children.dprel.bag
ArrayList<String> argChildrenDeprel = new ArrayList<String>();
for (FeaturizedToken child : argChildrenObjectList) {
argChildrenDeprel.add(child.getDeprel());
}
ArrayList<String> argChildrenDeprelBag = bag(argChildrenDeprel);
feat = arg.getPos() + buildString(argChildrenDeprelBag);
feats.add(feat);
}
// ---------- Bjorkelund et al, CoNLL2009 features. ----------
public void addBjorkelundSenseFeatures(int idx, ArrayList<String> feats) {
FeaturizedToken pred = getFeatureObject(idx);
FeaturizedToken predParent = getFeatureObject(pred.getParent());
ArrayList<Integer> predChildren = pred.getChildren();
// PredWord, PredPOS, PredDeprel, PredFeats
addBjorkelundGenericFeatures(idx, feats, "Pred");
// predParentWord, predParentPOS, predParentFeats
addBjorkelundPredParentFeatures(predParent, feats);
// ChildDepSet, ChildWordSet, ChildPOSSet
addBjorkelundPredChildFeatures(predChildren, feats);
// TODO: DepSubCat: the subcategorization frame of the predicate, e.g. OBJ+OPRD+SUB.
}
public void addBjorkelundPairFeatures(int pidx, int aidx, ArrayList<String> feats) {
/* From Bjorkelund et al, CoNLL 2009.
* Prefixes:
* Pred: the predicate
* PredParent: the parent of the predicate
* Arg: the argument
* Left: the leftmost dependent of the argument
* Right: the rightmost dependent of the argument
* LeftSibling: the left sibling of the argument
* RightSibling: the right sibling of the argument
*/
String feat;
FeaturizedToken pred = getFeatureObject(pidx);
FeaturizedToken arg = getFeatureObject(aidx);
FeaturizedTokenPair predArgPair = getFeatureObject(pidx, aidx);
FeaturizedToken predParent = getFeatureObject(pred.getParent());
ArrayList<Integer> predChildren = pred.getChildren();
FeaturizedToken argLeftSibling = getFeatureObject(arg.getNearLeftSibling());
FeaturizedToken argRightSibling = getFeatureObject(arg.getNearRightSibling());
FeaturizedToken argLeftDependent = getFeatureObject(arg.getFarLeftChild());
FeaturizedToken argRightDependent = getFeatureObject(arg.getFarRightChild());
// PredWord, PredPOS, PredFeats, PredDeprel
addBjorkelundGenericFeatures(pidx, feats, "Pred");
// ArgWord, ArgPOS, ArgFeats, ArgDeprel
addBjorkelundGenericFeatures(aidx, feats, "Arg");
// PredParentWord, PredParentPOS, PredParentFeats
addBjorkelundPredParentFeatures(predParent, feats);
// ChildDepSet, ChildWordSet, ChildPOSSet
addBjorkelundPredChildFeatures(predChildren, feats);
// LeftWord, LeftPOS, LeftFeats
addBjorkelundDependentFeats(argLeftDependent, feats, "Left");
// RightWord, RightPOS, RightFeats
addBjorkelundDependentFeats(argRightDependent, feats, "Right");
// LeftSiblingWord, LeftSiblingPOS, LeftSiblingFeats
addBjorkelundSiblingFeats(argLeftSibling, feats, "Left");
// RightSiblingWord, RightSiblingPOS, RightSiblingFeats
addBjorkelundSiblingFeats(argRightSibling, feats, "Right");
// DeprelPath, POSPath
addBjorkelundPathFeats(predArgPair, feats);
// Position
addBjorkelundPositionFeat(pidx, aidx, feats);
// PredLemma
feat = pred.getLemma();
feats.add("PredLemma:" + feat);
// TODO: Sense: the value of the Pred column, e.g. plan.01.
// TODO: DepSubCat: the subcategorization frame of the predicate, e.g. OBJ+OPRD+SUB.
}
public void addZhaoPairFeatures(int pidx, int aidx, ArrayList<String> feats) {
/* NOTE: Not sure about they're "Semantic Connection" features.
* What do they correspond to in CoNLL data?
* From paper: "This includes semantic head (semhead), left(right) farthest(nearest)
* semantic child (semlm, semln, semrm, semrn).
* We say a predicate is its argument's semantic head, and the latter is the former's child.
* Features related to this type may track the current semantic parsing status."
* If a semantic predicate is given, then the SRL task is moot... */
FeaturizedToken pred = getFeatureObject(pidx);
FeaturizedToken arg = getFeatureObject(aidx);
FeaturizedTokenPair predArgPair = getFeatureObject(pidx, aidx);
FeaturizedToken predLast = getFeatureObject(pidx - 1);
FeaturizedToken predNext = getFeatureObject(pidx + 1);
// FeatureObject predParent = getFeatureObject(pred.getParent());
FeaturizedToken argLast = getFeatureObject(aidx - 1);
FeaturizedToken argNext = getFeatureObject(aidx + 1);
FeaturizedToken argParent = getFeatureObject(arg.getParent());
ArrayList<Integer> predChildren = pred.getChildren();
ArrayList<Integer> argChildren = arg.getChildren();
List<Pair<Integer, Dir>> dependencyPath = predArgPair.getDependencyPath();
// Initialize Path structures.
// List<Pair<Integer, Dir>> dpPathPred = predArgPair.getDpPathPred();
List<Pair<Integer, Dir>> dpPathArg = predArgPair.getDpPathArg();
ArrayList<Integer> linePath = predArgPair.getLinePath();
ArrayList<FeaturizedToken> predChildrenObjectList = getFeatureObjectList(predChildren);
ArrayList<FeaturizedToken> argChildrenObjectList = getFeatureObjectList(argChildren);
ArrayList<FeaturizedToken> dependencyPathObjectList = getFeatureObjectList(dependencyPath);
ArrayList<FeaturizedToken> linePathCoNLL = getFeatureObjectList(linePath);
// Add the supervised features
if (prm.withSupervision) {
addZhaoSupervisedPredFeats(pred, arg, predLast, predNext, predChildrenObjectList, feats);
addZhaoSupervisedArgFeats(
pred, arg, predLast, predNext, argLast, argNext, argParent, argChildrenObjectList, feats);
addZhaoSupervisedCombinedFeats(
pred, arg, feats, dependencyPathObjectList, linePathCoNLL, dpPathArg);
}
// Add the unsupervised features
addZhaoUnsupervisedPredFeats(pred, predLast, predNext, feats);
addZhaoUnsupervisedArgFeats(arg, argLast, argNext, argChildrenObjectList, feats);
addZhaoUnsupervisedCombinedFeats(linePath, linePathCoNLL, feats);
}
private void getSecondThirdSupArgFeats(
FeaturizedToken pred,
FeaturizedToken arg,
FeaturizedToken predLast,
FeaturizedToken predNext,
FeaturizedToken argLast,
FeaturizedToken argNext,
FeaturizedToken argParent,
ArrayList<FeaturizedToken> argChildrenObjectList,
ArrayList<String> feats) {
FeaturizedToken argLm = getFeatureObject(arg.getFarLeftChild());
FeaturizedToken argRm = getFeatureObject(arg.getFarRightChild());
FeaturizedToken argRn = getFeatureObject(arg.getNearRightChild());
// FeatureObject argLn = getFeatureObject(arg.getNearLeftChild());
String feat;
// a.h.lemma
feats.add(argParent.getLemma());
// a.lm.dprel + a.form
feats.add(argLm.getDeprel() + "_" + arg.getForm());
// a.lm_1.lemma
feats.add(getFeatureObject(argLast.getFarLeftChild()).getLemma());
// a.lmn.pos (n=0,1)
feats.add(argLm.getPos());
feats.add(getFeatureObject(argNext.getFarLeftChild()).getPos());
// a.noFarChildren.pos.bag + a.rm.form
ArrayList<Integer> noFarChildren = arg.getNoFarChildren();
ArrayList<String> noFarChildrenPos = new ArrayList<String>();
for (Integer i : noFarChildren) {
noFarChildrenPos.add(getFeatureObject(i).getPos());
}
ArrayList<String> argNoFarChildrenBag = bag(noFarChildrenPos);
feat = buildString(argNoFarChildrenBag) + argRm.getForm();
feats.add(feat);
// a.pphead.lemma
feats.add(getFeatureObject(arg.getParent()).getLemma());
// a.rm.dprel + a.form
feats.add(argRm.getDeprel() + "_" + arg.getForm());
// a.rm_1.form
feats.add(getFeatureObject(argLast.getFarRightChild()).getForm());
// a.rm.lemma
feats.add(argRm.getLemma());
// a.rn.dprel + a.form
feats.add(argRn.getDeprel() + "_" + arg.getForm());
}