private LabeledSentence generateSupersenseTaggingInput(Tree sentence) {
LabeledSentence res = new LabeledSentence();
List<Tree> leaves = sentence.getLeaves();
for (int i = 0; i < leaves.size(); i++) {
String word = leaves.get(i).label().toString();
Tree preterm = leaves.get(i).parent(sentence);
String pos = preterm.label().toString();
String stem = AnalysisUtilities.getInstance().getLemma(word, pos);
res.addToken(word, stem, pos, "0");
}
return res;
}
public List<String> annotateSentenceWithSupersenses(Tree sentence) {
List<String> result = new ArrayList<String>();
int numleaves = sentence.getLeaves().size();
if (numleaves <= 1) {
return result;
}
LabeledSentence labeled = generateSupersenseTaggingInput(sentence);
// see if a NER socket server is available
int port = new Integer(ARKref.getProperties().getProperty("supersenseServerPort", "5557"));
String host = "127.0.0.1";
Socket client;
PrintWriter pw;
BufferedReader br;
String line;
try {
client = new Socket(host, port);
pw = new PrintWriter(client.getOutputStream());
br = new BufferedReader(new InputStreamReader(client.getInputStream()));
String inputStr = "";
for (int i = 0; i < labeled.length(); i++) {
String token = labeled.getTokens().get(i);
String stem = labeled.getStems().get(i);
String pos = labeled.getPOS().get(i);
inputStr += token + "\t" + stem + "\t" + pos + "\n";
}
pw.println(inputStr);
pw.flush(); // flush to complete the transmission
while ((line = br.readLine()) != null) {
String[] parts = line.split("\\t");
result.add(parts[2]);
}
br.close();
pw.close();
client.close();
} catch (Exception ex) {
if (ARKref.Opts.debug) System.err.println("Could not connect to SST server.");
// ex.printStackTrace();
}
// if socket server not available, then use a local NER object
if (result.size() == 0) {
try {
if (sst == null) {
DiscriminativeTagger.loadProperties(ARKref.getPropertiesPath());
sst =
DiscriminativeTagger.loadModel(
ARKref.getProperties()
.getProperty("supersenseModelFile", "config/supersenseModel.ser.gz"));
}
sst.findBestLabelSequenceViterbi(labeled, sst.getWeights());
for (String pred : labeled.getPredictions()) {
result.add(pred);
}
} catch (Exception e) {
e.printStackTrace();
}
}
// add a bunch of blanks if necessary
while (result.size() < numleaves) result.add("0");
if (ARKref.Opts.debug) System.err.println("annotateSentenceSST: " + result);
return result;
}