public void createInstances(String file, Instances is) throws Exception {
CONLLReader09 depReader = new CONLLReader09(file);
mf.register(REL, "<root-type>");
// register at least one predicate since the parsing data might not contain predicates as in
// the Japaness corpus but the development sets contains some
long sl = 0;
System.out.print("Registering feature parts of sentence: ");
int ic = 0;
int del = 0;
while (true) {
SentenceData09 instance = depReader.getNext();
if (instance == null) break;
ic++;
sl += instance.labels.length;
if (ic % 1000 == 0) {
del = outValue(ic, del);
}
String[] labs1 = instance.labels;
for (int i1 = 0; i1 < labs1.length; i1++) mf.register(REL, labs1[i1]);
String[] w = instance.forms;
for (int i1 = 0; i1 < w.length; i1++) mf.register(WORD, depReader.normalize(w[i1]));
w = instance.plemmas;
for (int i1 = 0; i1 < w.length; i1++) mf.register(WORD, depReader.normalize(w[i1]));
w = instance.ppos;
for (int i1 = 0; i1 < w.length; i1++) mf.register(POS, w[i1]);
w = instance.gpos;
for (int i1 = 0; i1 < w.length; i1++) mf.register(POS, w[i1]);
if (instance.feats != null) {
String fs[][] = instance.feats;
for (int i1 = 0; i1 < fs.length; i1++) {
w = fs[i1];
if (w == null) continue;
for (int i2 = 0; i2 < w.length; i2++) mf.register(FEAT, w[i2]);
}
}
if ((ic - 1) > options.count) break;
}
del = outValue(ic, del);
System.out.println();
Extractor.initFeatures();
Extractor.maxForm = mf.getFeatureCounter().get(WORD);
if (options.clusterFile == null) cl = new Cluster();
else cl = new Cluster(options.clusterFile, mf, 6);
mf.calculateBits();
Extractor.initStat(options.featureCreation);
System.out.println("" + mf.toString());
for (Extractor e : extractor) e.init();
depReader.startReading(file);
int num1 = 0;
is.init(ic, new MFO());
Edges.init(mf.getFeatureCounter().get(POS));
System.out.print("Creating edge filters and read corpus: ");
del = 0;
while (true) {
if (num1 % 100 == 0) del = outValue(num1, del);
SentenceData09 instance1 = depReader.getNext(is);
if (instance1 == null) break;
int last = is.size() - 1;
short[] pos = is.pposs[last];
for (int k = 0; k < is.length(last); k++) {
if (is.heads[last][k] < 0) continue;
Edges.put(pos[is.heads[last][k]], pos[k], is.labels[last][k]);
// Edges.put(pos[k],pos[is.heads[last][k]], is.labels[last][k]);
}
if (!options.allFeatures && num1 > options.count) break;
num1++;
}
del = outValue(num1, del);
System.out.println();
Edges.findDefault();
}
public void createInstances(String file, Instances is)
// throws Exception
{
CONLLReader09 depReader = new CONLLReader09(file);
mf.register(REL, "<root-type>");
// register at least one predicate since the parsing data might not contain predicates as in
// the Japaness corpus but the development sets contains some
long sl = 0;
System.out.print("Registering feature parts of sentence: ");
int ic = 0;
int del = 0;
while (true) {
SentenceData09 instance = depReader.getNext();
if (instance == null) break;
ic++;
sl += instance.labels.length;
if (ic % 1000 == 0) {
del = outValue(ic, del);
}
String[] labs1 = instance.labels;
for (int i1 = 0; i1 < labs1.length; i1++) mf.register(REL, labs1[i1]);
String[] w = instance.forms;
for (int i1 = 0; i1 < w.length; i1++) mf.register(WORD, depReader.normalize(w[i1]));
w = instance.plemmas;
for (int i1 = 0; i1 < w.length; i1++) mf.register(WORD, depReader.normalize(w[i1]));
w = instance.ppos;
for (int i1 = 0; i1 < w.length; i1++) mf.register(POS, w[i1]);
w = instance.gpos;
for (int i1 = 0; i1 < w.length; i1++) mf.register(POS, w[i1]);
if (instance.feats != null) {
String fs[][] = instance.feats;
for (int i1 = 0; i1 < fs.length; i1++) {
w = fs[i1];
if (w == null) continue;
for (int i2 = 0; i2 < w.length; i2++) mf.register(FEAT, w[i2]);
}
}
if ((ic - 1) > options.count) break;
}
del = outValue(ic, del);
System.out.println();
ExtractorReranker.initFeatures();
ExtractorReranker.maxForm = mf.getFeatureCounter().get(WORD);
if (options.clusterFile == null) cl = new Cluster();
else cl = new Cluster(options.clusterFile, mf, 6);
mf.calculateBits();
extractor.initStat();
System.out.println("" + mf.toString());
extractor.init();
depReader.startReading(file);
int num1 = 0;
is.init(ic, new MFB());
Edges.init(mf.getFeatureCounter().get(POS));
del = 0;
del = outValue(num1, del);
System.out.println();
}