Examples with FeatureResult com.joliciel.talismane.machineLearning.features.FeatureResult used on opensource projects

Example 16 with FeatureResult

use of com.joliciel.talismane.machineLearning.features.FeatureResult in project talismane by joliciel-informatique.

the class LinearSVMModelTrainer method getFeatureMatrix.

private Feature[][] getFeatureMatrix(ClassificationEventStream corpusEventStream, TObjectIntMap<String> featureIndexMap, TObjectIntMap<String> outcomeIndexMap, TIntList outcomeList, TIntIntMap featureCountMap, CountingInfo countingInfo) {
    try {
        int maxFeatureCount = 0;
        List<Feature[]> fullFeatureList = new ArrayList<Feature[]>();
        while (corpusEventStream.hasNext()) {
            ClassificationEvent corpusEvent = corpusEventStream.next();
            int outcomeIndex = outcomeIndexMap.get(corpusEvent.getClassification());
            if (outcomeIndex < 0) {
                outcomeIndex = countingInfo.currentOutcomeIndex++;
                outcomeIndexMap.put(corpusEvent.getClassification(), outcomeIndex);
            }
            outcomeList.add(outcomeIndex);
            Map<Integer, Feature> featureList = new TreeMap<Integer, Feature>();
            for (FeatureResult<?> featureResult : corpusEvent.getFeatureResults()) {
                if (featureResult.getOutcome() instanceof List) {
                    @SuppressWarnings("unchecked") FeatureResult<List<WeightedOutcome<String>>> stringCollectionResult = (FeatureResult<List<WeightedOutcome<String>>>) featureResult;
                    for (WeightedOutcome<String> stringOutcome : stringCollectionResult.getOutcome()) {
                        String featureName = featureResult.getTrainingName() + "|" + featureResult.getTrainingOutcome(stringOutcome.getOutcome());
                        double value = stringOutcome.getWeight();
                        this.addFeatureResult(featureName, value, featureList, featureIndexMap, featureCountMap, countingInfo);
                    }
                } else {
                    double value = 1.0;
                    if (featureResult.getOutcome() instanceof Double) {
                        @SuppressWarnings("unchecked") FeatureResult<Double> doubleResult = (FeatureResult<Double>) featureResult;
                        value = doubleResult.getOutcome().doubleValue();
                    }
                    this.addFeatureResult(featureResult.getTrainingName(), value, featureList, featureIndexMap, featureCountMap, countingInfo);
                }
            }
            if (featureList.size() > maxFeatureCount)
                maxFeatureCount = featureList.size();
            // convert to array immediately, to avoid double storage
            int j = 0;
            Feature[] featureArray = new Feature[featureList.size()];
            for (Feature feature : featureList.values()) {
                featureArray[j] = feature;
                j++;
            }
            fullFeatureList.add(featureArray);
            countingInfo.numEvents++;
            if (countingInfo.numEvents % 1000 == 0) {
                LOG.debug("Processed " + countingInfo.numEvents + " events.");
            }
        }
        Feature[][] featureMatrix = new Feature[countingInfo.numEvents][];
        int i = 0;
        for (Feature[] featureArray : fullFeatureList) {
            featureMatrix[i] = featureArray;
            i++;
        }
        fullFeatureList = null;
        LOG.debug("Event count: " + countingInfo.numEvents);
        LOG.debug("Feature count: " + featureIndexMap.size());
        return featureMatrix;
    } catch (TalismaneException e) {
        LOG.error(e.getMessage(), e);
        throw new RuntimeException(e);
    } catch (IOException e) {
        LOG.error(e.getMessage(), e);
        throw new RuntimeException(e);
    }
}

Example 17 with FeatureResult

use of com.joliciel.talismane.machineLearning.features.FeatureResult in project talismane by joliciel-informatique.

the class PatternEventStream method next.

@Override
public ClassificationEvent next() throws TalismaneException, IOException {
    ClassificationEvent event = null;
    if (this.hasNext()) {
        TokenPatternMatch tokenPatternMatch = currentPatternMatches.get(currentIndex);
        TokeniserOutcome outcome = currentOutcomes.get(currentIndex);
        String classification = outcome.name();
        LOG.debug("next event, pattern match: " + tokenPatternMatch.toString() + ", outcome:" + classification);
        List<FeatureResult<?>> tokenFeatureResults = new ArrayList<FeatureResult<?>>();
        for (TokenPatternMatchFeature<?> feature : tokenPatternMatchFeatures) {
            RuntimeEnvironment env = new RuntimeEnvironment();
            FeatureResult<?> featureResult = feature.check(tokenPatternMatch, env);
            if (featureResult != null) {
                tokenFeatureResults.add(featureResult);
            }
        }
        if (LOG.isTraceEnabled()) {
            SortedSet<String> featureResultSet = tokenFeatureResults.stream().map(f -> f.toString()).collect(Collectors.toCollection(() -> new TreeSet<String>()));
            for (String featureResultString : featureResultSet) {
                LOG.trace(featureResultString);
            }
        }
        event = new ClassificationEvent(tokenFeatureResults, classification);
        currentIndex++;
        if (currentIndex == currentPatternMatches.size()) {
            currentPatternMatches = null;
        }
    }
    return event;
}

Example 18 with FeatureResult

use of com.joliciel.talismane.machineLearning.features.FeatureResult in project talismane by joliciel-informatique.

the class PerceptronDetailedAnalysisWriter method onAnalyse.

/*
   * (non-Javadoc)
   * 
   * @see com.joliciel.talismane.maxent.MaxentObserver#onAnalyse(java.util.List,
   * java.util.Collection)
   */
@Override
public void onAnalyse(Object event, List<FeatureResult<?>> featureResults, Collection<Decision> decisions) throws IOException {
    Map<String, Double> outcomeTotals = new TreeMap<String, Double>();
    for (String outcome : modelParams.getOutcomes()) outcomeTotals.put(outcome, 0.0);
    writer.append("####### Event: " + event.toString() + "\n");
    writer.append("### Feature results:\n");
    for (FeatureResult<?> featureResult : featureResults) {
        if (featureResult.getOutcome() instanceof List) {
            @SuppressWarnings("unchecked") FeatureResult<List<WeightedOutcome<String>>> stringCollectionResult = (FeatureResult<List<WeightedOutcome<String>>>) featureResult;
            for (WeightedOutcome<String> stringOutcome : stringCollectionResult.getOutcome()) {
                String featureName = featureResult.getTrainingName() + "|" + featureResult.getTrainingOutcome(stringOutcome.getOutcome());
                String featureOutcome = stringOutcome.getOutcome();
                double value = stringOutcome.getWeight();
                this.writeFeatureResult(featureName, featureOutcome, value, outcomeTotals);
            }
        } else {
            double value = 1.0;
            if (featureResult.getFeature() instanceof DoubleFeature) {
                value = (Double) featureResult.getOutcome();
            }
            this.writeFeatureResult(featureResult.getTrainingName(), featureResult.getOutcome().toString(), value, outcomeTotals);
        }
    }
    List<Integer> featureIndexList = new ArrayList<Integer>();
    List<Double> featureValueList = new ArrayList<Double>();
    modelParams.prepareData(featureResults, featureIndexList, featureValueList);
    double[] results = decisionMaker.predict(featureIndexList, featureValueList);
    writer.append("### Outcome totals:\n");
    writer.append(String.format("%1$-30s", "outcome") + String.format("%1$#15s", "total") + String.format("%1$#15s", "normalised") + "\n");
    int j = 0;
    for (String outcome : modelParams.getOutcomes()) {
        double total = outcomeTotals.get(outcome);
        double normalised = results[j++];
        writer.append(String.format("%1$-30s", outcome) + String.format("%1$#15s", decFormat.format(total)) + String.format("%1$#15s", decFormat.format(normalised)) + "\n");
    }
    writer.append("\n");
    Map<String, Double> outcomeWeights = new TreeMap<String, Double>();
    for (Decision decision : decisions) {
        outcomeWeights.put(decision.getOutcome(), decision.getProbability());
    }
    writer.append("### Outcome list:\n");
    Set<WeightedOutcome<String>> weightedOutcomes = new TreeSet<WeightedOutcome<String>>();
    for (String outcome : modelParams.getOutcomes()) {
        Double weightObj = outcomeWeights.get(outcome);
        double weight = (weightObj == null ? 0.0 : weightObj.doubleValue());
        WeightedOutcome<String> weightedOutcome = new WeightedOutcome<String>(outcome, weight);
        weightedOutcomes.add(weightedOutcome);
    }
    for (WeightedOutcome<String> weightedOutcome : weightedOutcomes) {
        writer.append(String.format("%1$-30s", weightedOutcome.getOutcome()) + String.format("%1$#15s", decFormat.format(weightedOutcome.getWeight())) + "\n");
    }
    writer.append("\n");
    writer.flush();
}

Example 19 with FeatureResult

use of com.joliciel.talismane.machineLearning.features.FeatureResult in project jochre by urieli.

the class JochreLetterEventStream method next.

@Override
public ClassificationEvent next() {
    ClassificationEvent event = null;
    if (this.hasNext()) {
        Shape shape = shapeInSequence.getShape();
        LOG.debug("next event, shape: " + shape);
        LetterGuesserContext context = new LetterGuesserContext(shapeInSequence, history);
        List<FeatureResult<?>> featureResults = new ArrayList<>();
        // analyse features
        for (LetterFeature<?> feature : features) {
            RuntimeEnvironment env = new RuntimeEnvironment();
            FeatureResult<?> featureResult = feature.check(context, env);
            if (featureResult != null) {
                featureResults.add(featureResult);
                if (LOG.isTraceEnabled()) {
                    LOG.trace(featureResult.toString());
                }
            }
        }
        String outcome = shape.getLetter();
        event = new ClassificationEvent(featureResults, outcome);
        history.getLetters().add(outcome);
        // set shape to null so that hasNext can retrieve the next one.
        this.shapeInSequence = null;
    }
    return event;
}

Example 20 with FeatureResult

use of com.joliciel.talismane.machineLearning.features.FeatureResult in project jochre by urieli.

the class LetterGuesser method guessLetter.

/**
 * Analyses this shape, using the context provided for features that are not
 * intrinsic. Updates shape.getWeightedOutcomes to include all outcomes
 * above a certain threshold of probability.
 *
 * @return the best outcome for this shape.
 */
public String guessLetter(ShapeInSequence shapeInSequence, LetterSequence history) {
    Shape shape = shapeInSequence.getShape();
    if (LOG.isTraceEnabled())
        LOG.trace("guessLetter, shape: " + shape);
    List<FeatureResult<?>> featureResults = new ArrayList<FeatureResult<?>>();
    // analyse features
    for (LetterFeature<?> feature : features) {
        LetterGuesserContext context = new LetterGuesserContext(shapeInSequence, history);
        RuntimeEnvironment env = new RuntimeEnvironment();
        FeatureResult<?> featureResult = feature.check(context, env);
        if (featureResult != null) {
            featureResults.add(featureResult);
            if (LOG.isTraceEnabled()) {
                LOG.trace(featureResult.toString());
            }
        }
    }
    List<Decision> letterGuesses = decisionMaker.decide(featureResults);
    // store outcomes
    String bestOutcome = null;
    shape.getLetterGuesses().clear();
    for (Decision letterGuess : letterGuesses) {
        if (letterGuess.getProbability() >= MIN_PROB_TO_STORE) {
            shape.getLetterGuesses().add(letterGuess);
        }
    }
    bestOutcome = shape.getLetterGuesses().iterator().next().getOutcome();
    if (LOG.isTraceEnabled()) {
        LOG.trace("Shape: " + shape);
        LOG.trace("Letter: " + shape.getLetter());
        LOG.trace("Best outcome: " + bestOutcome);
    }
    return bestOutcome;
}