public void addSimplePairFeatures(int pidx, int aidx, ArrayList<String> feats) {
String predForm = sent.getWord(pidx);
String argForm = sent.getWord(aidx);
Set<String> a = sig.getSimpleUnkFeatures(predForm, pidx, cs.prm.language);
for (String c : a) {
feats.add(c);
}
Set<String> b = sig.getSimpleUnkFeatures(argForm, aidx, cs.prm.language);
for (String c : b) {
feats.add(c);
}
}
/** Adds factors and variables to the given factor graph. */
public void build(
SimpleAnnoSentence sent,
CorpusStatistics cs,
ObsFeatureExtractor obsFe,
ObsFeatureConjoiner ofc,
FactorGraph fg) {
build(
sent.getWords(),
sent.getLemmas(),
sent.getKnownPreds(),
cs.roleStateNames,
cs.predSenseListMap,
obsFe,
ofc,
fg);
}
// ---------- Meg's "Simple" features. ----------
public void addSimpleSoloFeatures(int idx, ArrayList<String> feats) {
String wordForm = sent.getWord(idx);
Set<String> a = sig.getSimpleUnkFeatures(wordForm, idx, cs.prm.language);
for (String c : a) {
feats.add(c);
}
}
// ---------- Naradowsky et al.'s 2011 ACL features. ----------
public void addNaradowskySenseFeatures(int idx, ArrayList<String> feats) {
String word = sent.getWord(idx);
String wordForm = decideForm(word, idx);
String wordPos = sent.getPosTag(idx);
String wordLemma = sent.getLemma(idx);
feats.add("head_" + wordForm + "_word");
feats.add("head_" + wordLemma + "_lemma");
feats.add("head_" + wordPos + "_tag");
feats.add("head_" + wordForm + "_" + wordPos + "_wordtag");
String cap;
if (FeaturizedToken.capitalized(wordForm)) {
cap = "UC";
} else {
cap = "LC";
}
feats.add("head_" + cap + "_caps");
}
public int getSentSize() {
return sent.size();
}
/**
* Made up by Meg based on pair and sense features. public void addNaradowskySoloFeatures(int idx,
* ArrayList<String> feats) { addNaradowskySenseFeatures(idx, feats); String word =
* sent.getWord(idx); String wordForm = decideForm(word, idx); String wordPos =
* sent.getPosTag(idx);
*
* <p>feats.add("head_" + wordForm + "_dep_" + wordPos + "_wordpos"); feats.add("slen_" +
* sent.size()); feats.add("head_" + wordForm + "_word"); feats.add("head_" + wordPos + "_tag"); }
* *
*/
public void addNaradowskyPairFeatures(int pidx, int aidx, ArrayList<String> feats) {
String predWord = sent.getWord(pidx);
String argWord = sent.getWord(aidx);
String predForm = decideForm(predWord, pidx);
String argForm = decideForm(argWord, aidx);
String predPos = sent.getPosTag(pidx);
String argPos = sent.getPosTag(aidx);
String dir;
int dist = Math.abs(aidx - pidx);
if (aidx > pidx) dir = "RIGHT";
else if (aidx < pidx) dir = "LEFT";
else dir = "SAME";
feats.add("head_" + predForm + "dep_" + argForm + "_word");
feats.add("head_" + predPos + "_dep_" + argPos + "_pos");
feats.add("head_" + predForm + "_dep_" + argPos + "_wordpos");
feats.add("head_" + predPos + "_dep_" + argForm + "_posword");
feats.add(
"head_"
+ predForm
+ "_dep_"
+ argForm
+ "_head_"
+ predPos
+ "_dep_"
+ argPos
+ "_wordwordpospos");
feats.add("head_" + predPos + "_dep_" + argPos + "_dist_" + dist + "_posdist");
feats.add("head_" + predPos + "_dep_" + argPos + "_dir_" + dir + "_posdir");
feats.add("head_" + predPos + "_dist_" + dist + "_dir_" + dir + "_posdistdir");
feats.add("head_" + argPos + "_dist_" + dist + "_dir_" + dir + "_posdistdir");
feats.add("slen_" + sent.size());
feats.add("dir_" + dir);
feats.add("dist_" + dist);
feats.add("dir_dist_" + dir + dist);
feats.add("head_" + predForm + "_word");
feats.add("head_" + predPos + "_tag");
feats.add("arg_" + argForm + "_word");
feats.add("arg_" + argPos + "_tag");
if (prm.withSupervision) {
List<String> predFeats = sent.getFeats(pidx);
List<String> argFeats = sent.getFeats(aidx);
if (predFeats == null) {
predFeats = new ArrayList<String>();
predFeats.add("_");
}
if (argFeats == null) {
argFeats = new ArrayList<String>();
argFeats.add("_");
}
for (String m1 : predFeats) {
for (String m2 : argFeats) {
feats.add(m1 + "_" + m2 + "_morph");
}
}
}
}