Examples with POSBracketToToken edu.illinois.cs.cogcomp.lbjava.nlp.seg.POSBracketToToken used on opensource projects

Example 1 with POSBracketToToken

use of edu.illinois.cs.cogcomp.lbjava.nlp.seg.POSBracketToToken in project cogcomp-nlp by CogComp.

the class POSTrain method trainModels.

/**
 * Trains the taggers with specified, labeled training data.
 *
 * @param trainingData The labeled training data
 */
public void trainModels(String trainingData) {
    // Set up the data
    Parser trainingParser = new POSBracketToToken(trainingData);
    Parser trainingParserUnknown = new POSLabeledUnknownWordParser(trainingData);
    MikheevTable.isTraining = true;
    BaselineTarget.isTraining = true;
    Object ex;
    // baseline and mikheev just count, they don't learn -- so one iteration should be enough
    while ((ex = trainingParser.next()) != null) {
        baselineTarget.learn(ex);
        mikheevTable.learn(ex);
    }
    baselineTarget.doneLearning();
    mikheevTable.doneLearning();
    trainingParser.reset();
    POSTaggerUnknown.isTraining = true;
    POSTaggerKnown.isTraining = true;
    // Run the learner
    for (int i = 0; i < iter; i++) {
        System.out.println("Training round " + i);
        while ((ex = trainingParser.next()) != null) {
            taggerKnown.learn(ex);
        }
        System.out.println("\tFinished training " + rm.getString("knownName"));
        while ((ex = trainingParserUnknown.next()) != null) {
            taggerUnknown.learn(ex);
        }
        System.out.println("\tFinished training " + rm.getString("unknownName"));
        trainingParser.reset();
        trainingParserUnknown.reset();
        taggerKnown.doneWithRound();
        taggerUnknown.doneWithRound();
    }
    taggerUnknown.doneLearning();
    taggerKnown.doneLearning();
}

Example 2 with POSBracketToToken

use of edu.illinois.cs.cogcomp.lbjava.nlp.seg.POSBracketToToken in project cogcomp-nlp by CogComp.

the class TestPOSModels method testAccuracy.

/**
 * Tags the unlabeled data and compares the part-of-speech tags with the labeled data, keeping
 * track of and reporting total accuracy at the end.
 */
public void testAccuracy() {
    WordForm __wordForm = new WordForm();
    Parser labeledParser = new POSBracketToToken(labeledTestFile);
    int numSeen = 0;
    int numEqual = 0;
    Token labeledWord = (Token) labeledParser.next();
    for (; labeledWord != null; labeledWord = (Token) labeledParser.next()) {
        String labeledTag = labeledWord.label;
        String testTag = tagger.discreteValue(labeledWord);
        if (labeledTag.equals(testTag)) {
            numEqual++;
        }
        numSeen++;
    }
    System.out.println("Total accuracy over " + numSeen + " items: " + String.format("%.2f", 100.0 * (double) numEqual / (double) numSeen) + "%");
}

Example 3 with POSBracketToToken

use of edu.illinois.cs.cogcomp.lbjava.nlp.seg.POSBracketToToken in project cogcomp-nlp by CogComp.

the class TestPOS method main.

/**
 * Implements the program described above.
 *
 * @param args The command line parameters.
 */
public static void main(String[] args) {
    // Parse the command line
    // if (args.length != 1) {
    // logger.error("usage: java edu.illinois.cs.cogcomp.lbj.pos.TestPOS <text file>");
    // System.exit(1);
    // }
    // String testingFile = args[0];
    ResourceManager rm = new POSConfigurator().getDefaultConfig();
    String testingFile = rm.getString("testData");
    TestDiscrete.testDiscrete(new TestDiscrete(), new POSTagger(), new POSLabel(), new POSBracketToToken(testingFile), true, 0);
}