@Override
protected AnalysisEngine[] createAnalysisEngines() throws ResourceInitializationException {
// Use OpenNlp to generate the POS etc for us
final ExternalResourceDescription tokensDesc =
ExternalResourceFactory.createExternalResourceDescription(
"tokens", SharedOpenNLPModel.class);
final ExternalResourceDescription sentencesDesc =
ExternalResourceFactory.createExternalResourceDescription(
"sentences", SharedOpenNLPModel.class);
final ExternalResourceDescription posDesc =
ExternalResourceFactory.createExternalResourceDescription(
"posTags", SharedOpenNLPModel.class);
final ExternalResourceDescription chunksDesc =
ExternalResourceFactory.createExternalResourceDescription(
"phraseChunks", SharedOpenNLPModel.class);
return asArray(
createAnalysisEngine(
OpenNLP.class,
"tokens",
tokensDesc,
"sentences",
sentencesDesc,
"posTags",
posDesc,
"phraseChunks",
chunksDesc),
createAnalysisEngine(MaltParser.class));
}
@Override
protected AnalysisEngine[] createAnalysisEngines() throws ResourceInitializationException {
final ExternalResourceDescription parserChunkingDesc =
ExternalResourceFactory.createExternalResourceDescription(
"parserChunking", SharedOpenNLPModel.class);
// Add in the OpenNLP implementation too, as its a prerequisite
// (in theory we should test OpenNLPParser in isolation, but in practise
// it as this as a
// dependency
// so better test they work together)
final ExternalResourceDescription tokensDesc =
ExternalResourceFactory.createExternalResourceDescription(
"tokens", SharedOpenNLPModel.class);
final ExternalResourceDescription sentencesDesc =
ExternalResourceFactory.createExternalResourceDescription(
"sentences", SharedOpenNLPModel.class);
final ExternalResourceDescription posDesc =
ExternalResourceFactory.createExternalResourceDescription(
"posTags", SharedOpenNLPModel.class);
final ExternalResourceDescription chunksDesc =
ExternalResourceFactory.createExternalResourceDescription(
"phraseChunks", SharedOpenNLPModel.class);
AnalysisEngineFactory.createEngineDescription();
return asArray(
createAnalysisEngine(
OpenNLP.class,
"tokens",
tokensDesc,
"sentences",
sentencesDesc,
"posTags",
posDesc,
"phraseChunks",
chunksDesc),
createAnalysisEngine(OpenNLPParser.class, "parserChunking", parserChunkingDesc));
}
public static AnalysisEngineDescription getDescription(String languageCode)
throws ResourceInitializationException {
String modelPath = String.format("/models/%s-pos-maxent.bin", languageCode);
return AnalysisEngineFactory.createEngineDescription(
PoStagger.class,
UimaUtil.MODEL_PARAMETER,
ExternalResourceFactory.createExternalResourceDescription(
POSModelResourceImpl.class, PoStagger.class.getResource(modelPath).toString()),
UimaUtil.SENTENCE_TYPE_PARAMETER,
Sentence.class.getName(),
UimaUtil.TOKEN_TYPE_PARAMETER,
Token.class.getName(),
UimaUtil.POS_FEATURE_PARAMETER,
"pos");
}