/*
* Model creation, saving, loading, and saving
*/
public void train(Annotation doc) {
List<List<CoreLabel>> trainingSet =
AnnotationUtils.entityMentionsToCoreLabels(doc, annotationsToSkip, useSubTypes, useBIO);
if (SAVE_CONLL_2003) {
// dump a file in CoNLL-2003 format
try {
PrintStream os = new PrintStream(new FileOutputStream("train.conll"));
// saveCoNLLFiles("/tmp/ace/train/", doc, useSubTypes, useBIO);
saveCoNLL(os, trainingSet, useBIO);
os.close();
} catch (IOException e) {
e.printStackTrace();
System.exit(1);
}
}
this.classifier = createClassifier();
if (trainingSet.size() > 0) {
this.classifier.train(Collections.unmodifiableCollection(trainingSet));
}
}
/**
* Annotate an ExtractionDataSet with entities. This will modify the ExtractionDataSet in place.
*
* @param doc The dataset to label
*/
@Override
public void annotate(Annotation doc) {
if (SAVE_CONLL_2003) {
// dump a file in CoNLL-2003 format
try {
PrintStream os = new PrintStream(new FileOutputStream("test.conll"));
List<List<CoreLabel>> labels =
AnnotationUtils.entityMentionsToCoreLabels(doc, annotationsToSkip, useSubTypes, useBIO);
BasicEntityExtractor.saveCoNLL(os, labels, true);
// saveCoNLLFiles("/tmp/ace/test", doc, useSubTypes, useBIO);
os.close();
} catch (IOException e) {
e.printStackTrace();
System.exit(1);
}
}
List<CoreMap> sents = doc.get(CoreAnnotations.SentencesAnnotation.class);
int sentCount = 1;
for (CoreMap sentence : sents) {
if (useNERTags) {
this.makeAnnotationFromAllNERTags(sentence);
} else extractEntities(sentence, sentCount);
sentCount++;
}
/*
if(SAVE_CONLL_2003){
try {
saveCoNLLFiles("test_output/", doc, useSubTypes, useBIO);
System.err.println("useBIO = " + useBIO);
} catch (IOException e) {
e.printStackTrace();
System.exit(1);
}
}
*/
}