public static Document fromTaggerDocumentToDocument(TaggerDocument doc) {
Document document = null;
if (doc != null) {
if (doc.getDoctype().equalsIgnoreCase(DocumentType.TWIITER_DOC)) {
document = new Tweet();
} else if (doc.getDoctype().equalsIgnoreCase(DocumentType.SMS_DOC)) {
document = new SMS();
} else if (doc.getDoctype().equalsIgnoreCase(DocumentType.FACEBOOK_DOC)) {
document = new Facebook();
}
document.setDocumentID(doc.getDocumentID());
document.setCrisisID(doc.getCrisisID());
document.humanLabelCount = (doc.hasHumanLabels() == false) ? 0 : 1;
document.setCrisisCode(doc.getCrisisCode());
document.setLanguage(doc.getLanguage());
WordSet wordSet = new WordSet();
String text = doc.getWordFeatures();
wordSet.addAll(FeatureExtractor.getWordsInStringWithBigrams(text, false));
document.addFeatureSet(wordSet);
document.setValueAsTrainingSample(doc.getValueAsTrainingSample());
/*
List<NominalLabelBC> labels = doc.getHumanLabels(NominalLabelBC.class);
if (!labels.isEmpty()) {
for (NominalLabelBC label : labels) {
document.addLabel(label);
}
}*/
}
return document;
}
