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;
}
public static TaggerDocument fromDocumentToTaggerDocument(Document doc) {
TaggerDocument document = new TaggerDocument();
if (doc != null) {
// NOTE: documentID needs to be set separately as Auto Generation ID from DB/Hibernate
// Now copy the remaining fields
document.setHasHumanLabels(doc.hasHumanLabels());
document.setCrisisID(doc.getCrisisID());
document.setCrisisCode(doc.getCrisisCode());
document.setReceivedAt(
new java.sql.Timestamp(java.util.Calendar.getInstance().getTimeInMillis()));
document.setLanguage(doc.getLanguage());
document.setDoctype(doc.getClass().getSimpleName().toString());
if (doc.getInputJson() != null) {
document.setData(Helpers.escapeJson(doc.getInputJson().toString()));
} else {
document.setData(null);
}
if (doc.features != null) {
document.setWordFeatures(DocumentJSONConverter.getFeaturesJson(WordSet.class, doc));
}
document.setGeoFeatures(null);
document.setValueAsTrainingSample(doc.getValueAsTrainingSample());
boolean val = Math.random() < (1.0 / 5.0) ? true : false;
document.setIsEvaluationSet(val);
/*
List<NominalLabelBC> labels = doc.getHumanLabels(NominalLabelBC.class);
if (!labels.isEmpty()) {
List<NominalLabel> nbList = new ArrayList<NominalLabel>();
for (NominalLabelBC label : labels) {
NominalLabel nb = new NominalLabel(label.getNominalLabelID());
nbList.add(nb);
}
document.setNominalLabelCollection(nbList);
} else {
document.setNominalLabelCollection(null);
}*/
return document;
}
return null;
}