@Override
public boolean equals(Object unit) {
if (!(unit instanceof WordLabelUnit)) {
return false;
}
WordLabelUnit labelUnit = (WordLabelUnit) unit;
if (this.isNegative() && labelUnit.isNegative()) {
return true;
}
return getLabel().equals(labelUnit.getLabel());
}
protected SparseFeatureVector calcWordGlobalFeatures(
Word word, Instance instance, Label y, int lastIndex, WordLabelUnit candidateLabelUnit) {
SparseFeatureVector globalFv = new SparseFeatureVector(params);
StringSparseVector wordPairFv = new StringSparseVector(params);
StringSparseVector wordFv = new StringSparseVector(params);
String candidateLabel = candidateLabelUnit.getLabel();
String wBase = word.getWord().getRealBase();
String wPOS = word.getWord().getPOS();
int[] contextWords = new int[4];
Arrays.fill(contextWords, -1);
for (int i = 0; i <= lastIndex; i++) {
if (y.getLabel(i) instanceof PairLabelUnit) {
if (y.getLabel(i).isNegative()) continue;
if (candidateLabelUnit.isNegative()) {
continue;
}
Pair pair = (Pair) instance.getSequence().get(i);
if (pair.getW1().getId() == word.getId()) {
String pairLabel = ((PairLabelUnit) y.getLabel(i)).getLabel();
wordPairFv.add(candidateLabel.concat(pairLabel), 1.);
wordPairFv.add(candidateLabel.concat(wBase).concat(pairLabel), 1.);
if (params.getUseFullFeatures()) {
// w-rel-w
int w2Idx = ((JointInstance) instance).getWord(pair.getW2().getId());
if (w2Idx <= lastIndex) {
String w2Base = pair.getW2().getWord().getRealBase();
String w2Label = ((WordLabelUnit) y.getLabel(w2Idx)).getLabel();
wordPairFv.add(candidateLabel.concat(pairLabel).concat(w2Label), 1.);
wordPairFv.add(
candidateLabel.concat(wBase).concat(pairLabel).concat(w2Label).concat(w2Base),
1.);
}
}
} else if (pair.getW2().getId() == word.getId()) {
String pairLabel = ((PairLabelUnit) y.getLabel(i)).getLabel();
wordPairFv.add(pairLabel.concat(candidateLabel), 1.);
wordPairFv.add(pairLabel.concat(wBase).concat(candidateLabel), 1.);
if (params.getUseFullFeatures()) {
// w-rel-w
int w1Idx = ((JointInstance) instance).getWord(pair.getW1().getId());
if (w1Idx <= lastIndex) {
String w1Base = pair.getW1().getWord().getRealBase();
String w1Label = ((WordLabelUnit) y.getLabel(w1Idx)).getLabel();
wordPairFv.add(w1Label.concat(pairLabel).concat(candidateLabel), 1.);
wordPairFv.add(
w1Base.concat(w1Label).concat(pairLabel).concat(candidateLabel).concat(wBase),
1.);
}
}
}
} else {
assert y.getLabel(i) instanceof WordLabelUnit;
if (!params.getUseGlobalEntityFeatures()) continue;
Word w2 = (Word) instance.getSequence().get(i);
if (w2.getId() == word.getId() - 2) {
contextWords[0] = i;
}
if (w2.getId() == word.getId() - 1) {
String wordLabel = ((WordLabelUnit) y.getLabel(i)).getLabel();
contextWords[1] = i;
addBigramToFV(
wordFv,
wordLabel,
w2.getWord().getRealBase(),
w2.getWord().getPOS(),
candidateLabel,
wBase,
wPOS);
} else if (word.getId() == w2.getId() - 1) {
String wordLabel = ((WordLabelUnit) y.getLabel(i)).getLabel();
contextWords[2] = i;
addBigramToFV(
wordFv,
candidateLabel,
wBase,
wPOS,
wordLabel,
w2.getWord().getRealBase(),
w2.getWord().getPOS());
} else if (word.getId() == w2.getId() - 2) {
contextWords[3] = i;
}
}
}
List<Word> entityWords = getEntityWords(word, instance, y, lastIndex, candidateLabelUnit);
if (entityWords.size() > 0) {
if (params.getUseFullFeatures()) {
String entString = entityString(entityWords);
String candidateLabelType = candidateLabelUnit.getType();
wordFv.add(candidateLabelType.concat(entString), 1.);
Word lastWord = entityWords.get(entityWords.size() - 1);
for (int i = 0; i <= lastIndex; i++) {
if (y.getLabel(i) instanceof PairLabelUnit) {
if (y.getLabel(i).isNegative()) continue;
Pair pair = (Pair) instance.getSequence().get(i);
if (pair.getW1().getId() == lastWord.getId()) {
String pairLabel = ((PairLabelUnit) y.getLabel(i)).getLabel();
wordPairFv.add(candidateLabelType.concat(pairLabel), 1.);
wordPairFv.add(candidateLabelType.concat(entString).concat(pairLabel), 1.);
// ent-rel-ent
int w2Idx = ((JointInstance) instance).getWord(pair.getW2().getId());
if (w2Idx <= lastIndex) {
List<Word> e2Words =
getEntityWords(
pair.getW2(), instance, y, lastIndex, (WordLabelUnit) y.getLabel(w2Idx));
if (e2Words.size() > 0) {
String e2String = entityString(e2Words);
String w2LabelType = ((WordLabelUnit) y.getLabel(w2Idx)).getType();
wordPairFv.add(candidateLabelType.concat(pairLabel).concat(w2LabelType), 1.);
wordPairFv.add(
candidateLabelType
.concat(entString)
.concat(pairLabel)
.concat(w2LabelType)
.concat(e2String),
1.);
}
}
} else if (pair.getW2().getId() == lastWord.getId()) {
String pairLabel = ((PairLabelUnit) y.getLabel(i)).getLabel();
wordPairFv.add(pairLabel.concat(candidateLabelType), 1.);
wordPairFv.add(pairLabel.concat(candidateLabelType).concat(entString), 1.);
// ent-rel-ent
int w1Idx = ((JointInstance) instance).getWord(pair.getW1().getId());
if (w1Idx <= lastIndex) {
List<Word> e1Words =
getEntityWords(
pair.getW1(), instance, y, lastIndex, (WordLabelUnit) y.getLabel(w1Idx));
if (e1Words.size() > 0) {
String e1String = entityString(e1Words);
String w1LabelType = ((WordLabelUnit) y.getLabel(w1Idx)).getType();
wordPairFv.add(w1LabelType.concat(pairLabel).concat(candidateLabelType), 1.);
wordPairFv.add(
w1LabelType
.concat(e1String)
.concat(pairLabel)
.concat(candidateLabelType)
.concat(entString),
1.);
}
}
}
}
}
}
}
if (contextWords[0] >= 0 && contextWords[1] >= 0) {
Word w0 = (Word) instance.getSequence().get(contextWords[0]);
String l0 = ((WordLabelUnit) y.getLabel(contextWords[0])).getLabel();
Word w1 = (Word) instance.getSequence().get(contextWords[1]);
String l1 = ((WordLabelUnit) y.getLabel(contextWords[1])).getLabel();
addTrigramToFV(
wordFv,
l0,
w0.getWord().getRealBase(),
w0.getWord().getPOS(),
l1,
w1.getWord().getRealBase(),
w1.getWord().getPOS(),
candidateLabel,
wBase,
wPOS);
}
if (contextWords[1] >= 0 && contextWords[2] >= 0) {
Word w1 = (Word) instance.getSequence().get(contextWords[1]);
String l1 = ((WordLabelUnit) y.getLabel(contextWords[1])).getLabel();
Word w2 = (Word) instance.getSequence().get(contextWords[2]);
String l2 = ((WordLabelUnit) y.getLabel(contextWords[2])).getLabel();
addTrigramToFV(
wordFv,
l1,
w1.getWord().getRealBase(),
w1.getWord().getPOS(),
candidateLabel,
wBase,
wPOS,
l2,
w2.getWord().getRealBase(),
w2.getWord().getPOS());
}
if (contextWords[2] >= 0 && contextWords[3] >= 0) {
Word w2 = (Word) instance.getSequence().get(contextWords[2]);
String l2 = ((WordLabelUnit) y.getLabel(contextWords[2])).getLabel();
Word w3 = (Word) instance.getSequence().get(contextWords[3]);
String l3 = ((WordLabelUnit) y.getLabel(contextWords[3])).getLabel();
addTrigramToFV(
wordFv,
candidateLabel,
wBase,
wPOS,
l2,
w2.getWord().getRealBase(),
w2.getWord().getPOS(),
l3,
w3.getWord().getRealBase(),
w3.getWord().getPOS());
}
globalFv.add(wordFv, "WORD");
wordPairFv.mult(params.getRelWeight());
globalFv.add(wordPairFv, "WORDPAIR");
return globalFv;
}
protected List<Word> getEntityWords(
Word word, Instance instance, Label y, int lastIndex, WordLabelUnit candidateLabelUnit) {
List<Word> words = new Vector<Word>();
if (candidateLabelUnit.isNegative()) {
return words;
}
int targetId = word.getId();
if (candidateLabelUnit.getPosition().equals("U")) {
words.add(word);
} else if (candidateLabelUnit.getPosition().equals("L")) {
words.add(word);
boolean found = false;
for (int i = targetId - 1; i >= 0; --i) {
int seqIndex = ((JointInstance) instance).getWord(i);
if (seqIndex >= lastIndex) {
break;
}
String pos = ((WordLabelUnit) y.getLabel(seqIndex)).getPosition();
if (pos.equals("B")) {
words.add(0, word);
found = true;
} else if (pos.equals("I")) {
words.add(0, word);
}
}
if (!found) {
words.clear();
}
} else if (candidateLabelUnit.getPosition().equals("I")) {
words.add(word);
boolean leftFound = false;
for (int i = targetId - 1; i >= 0; --i) {
int seqIndex = ((JointInstance) instance).getWord(i);
if (seqIndex >= lastIndex) {
break;
}
String pos = ((WordLabelUnit) y.getLabel(seqIndex)).getPosition();
if (pos.equals("B")) {
words.add(0, word);
leftFound = true;
} else if (pos.equals("I")) {
words.add(0, word);
}
}
if (!leftFound) {
words.clear();
} else {
boolean rightFound = false;
int nwords = ((JointInstance) instance).getNumWords();
for (int i = targetId + 1; i < nwords; ++i) {
int seqIndex = ((JointInstance) instance).getWord(i);
if (seqIndex >= lastIndex) {
break;
}
String pos = ((WordLabelUnit) y.getLabel(seqIndex)).getPosition();
if (pos.equals("L")) {
words.add(word);
rightFound = true;
} else if (pos.equals("I")) {
words.add(word);
}
}
if (!rightFound) {
words.clear();
}
}
} else if (candidateLabelUnit.getPosition().equals("B")) {
boolean found = false;
int nwords = ((JointInstance) instance).getNumWords();
for (int i = targetId + 1; i < nwords; ++i) {
int seqIndex = ((JointInstance) instance).getWord(i);
if (seqIndex >= lastIndex) {
break;
}
String pos = ((WordLabelUnit) y.getLabel(seqIndex)).getPosition();
if (pos.equals("L")) {
words.add(word);
found = true;
} else if (pos.equals("I")) {
words.add(word);
}
}
if (!found) {
words.clear();
}
}
return words;
}