public AnalysisEngineDescription getReplacementNormalization(
String aModelLocation, SrcSurroundings aSrc, TargetSurroundings aTarget)
throws ResourceInitializationException {
AggregateBuilder ab = new AggregateBuilder();
ab.add(
createEngineDescription(
ReplacementFileNormalizer.class,
ReplacementFileNormalizer.PARAM_MODEL_LOCATION,
aModelLocation,
ReplacementFileNormalizer.PARAM_SRC_SURROUNDINGS,
aSrc,
ReplacementFileNormalizer.PARAM_TARGET_SURROUNDINGS,
aTarget),
CAS.NAME_DEFAULT_SOFA,
getSourceView());
ab.add(
createEngineDescription(ApplyChangesAnnotator.class),
ApplyChangesAnnotator.VIEW_SOURCE,
getSourceView(),
ApplyChangesAnnotator.VIEW_TARGET,
getTargetView());
AnalysisEngineDescription aed = ab.createAggregateDescription();
// aed.setAnnotatorImplementationName(new File(filepath).getName().split("\\")[0]);
return aed;
}
public AnalysisEngineDescription getUmlautSharpSNormalization(
ExternalResourceDescription aFrequencyProvider, int aMinFrequency)
throws ResourceInitializationException {
AggregateBuilder ab = new AggregateBuilder();
ab.add(
createEngineDescription(BreakIteratorSegmenter.class),
CAS.NAME_DEFAULT_SOFA,
getSourceView());
ab.add(
createEngineDescription(
SharpSNormalizer.class,
SharpSNormalizer.FREQUENCY_PROVIDER,
aFrequencyProvider,
SharpSNormalizer.PARAM_MIN_FREQUENCY_THRESHOLD,
aMinFrequency),
CAS.NAME_DEFAULT_SOFA,
getSourceView());
ab.add(
createEngineDescription(ApplyChangesAnnotator.class),
ApplyChangesAnnotator.VIEW_SOURCE,
getSourceView(),
ApplyChangesAnnotator.VIEW_TARGET,
getTargetView());
AnalysisEngineDescription aed = ab.createAggregateDescription();
aed.setAnnotatorImplementationName("Umlaute");
return aed;
}
public AnalysisEngineDescription getCapitalizationNormalization(
ExternalResourceDescription aFrequencyProvider) throws ResourceInitializationException {
AggregateBuilder ab = new AggregateBuilder();
ab.add(
createEngineDescription(BreakIteratorSegmenter.class),
CAS.NAME_DEFAULT_SOFA,
getSourceView());
ab.add(
createEngineDescription(
CapitalizationNormalizer.class,
CapitalizationNormalizer.FREQUENCY_PROVIDER,
aFrequencyProvider),
CAS.NAME_DEFAULT_SOFA,
getSourceView());
ab.add(
createEngineDescription(ApplyChangesAnnotator.class),
ApplyChangesAnnotator.VIEW_SOURCE,
getSourceView(),
ApplyChangesAnnotator.VIEW_TARGET,
getTargetView());
AnalysisEngineDescription aed = ab.createAggregateDescription();
aed.setAnnotatorImplementationName("Capitalization");
return aed;
}
public AnalysisEngineDescription getSpellcorrection(String aModelLocation)
throws ResourceInitializationException {
AggregateBuilder ab = new AggregateBuilder();
ab.add(
createEngineDescription(BreakIteratorSegmenter.class),
CAS.NAME_DEFAULT_SOFA,
getSourceView());
ab.add(
createEngineDescription(
JazzyChecker.class, JazzyChecker.PARAM_MODEL_LOCATION, aModelLocation),
CAS.NAME_DEFAULT_SOFA,
getSourceView());
ab.add(
createEngineDescription(SpellingNormalizer.class), CAS.NAME_DEFAULT_SOFA, getSourceView());
ab.add(
createEngineDescription(ApplyChangesAnnotator.class),
ApplyChangesAnnotator.VIEW_SOURCE,
getSourceView(),
ApplyChangesAnnotator.VIEW_TARGET,
getTargetView());
AnalysisEngineDescription aed = ab.createAggregateDescription();
aed.setAnnotatorImplementationName("Spell");
return aed;
}
public static void main(String[] args) throws Exception {
Options options = CliFactory.parseArguments(Options.class, args);
List<File> testFiles =
DocumentClassificationEvaluation.getFilesFromDirectory(options.getTestDirectory());
DocumentClassificationEvaluation evaluation =
new DocumentClassificationEvaluation(options.getModelsDirectory());
CollectionReader collectionReader = evaluation.getCollectionReader(testFiles);
AggregateBuilder builder =
DocumentClassificationEvaluation.createDocumentClassificationAggregate(
options.getModelsDirectory(), AnnotatorMode.CLASSIFY);
SimplePipeline.runPipeline(
collectionReader,
builder.createAggregateDescription(),
AnalysisEngineFactory.createEngineDescription(PrintClassificationsAnnotator.class));
}