/** * Tokenize, emulating the Penn Treebank */ public Annotator tokenizer(Properties properties) { return new TokenizerAnnotator(properties); }
private static AnnotationPipeline makeNumericPipeline() { AnnotationPipeline pipeline = new AnnotationPipeline(); pipeline.addAnnotator(new TokenizerAnnotator(false, "en")); pipeline.addAnnotator(new WordsToSentencesAnnotator(false)); pipeline.addAnnotator(new POSTaggerAnnotator(false)); pipeline.addAnnotator(new TimeAnnotator(true)); return pipeline; }
public static AnnotationPipeline getPipeline(Properties props, boolean tokenize) throws Exception { // useGUTime = Boolean.parseBoolean(props.getProperty("gutime", "false")); AnnotationPipeline pipeline = new AnnotationPipeline(); if (tokenize) { pipeline.addAnnotator(new TokenizerAnnotator(false, "en")); pipeline.addAnnotator(new WordsToSentencesAnnotator(false)); } pipeline.addAnnotator(new POSTaggerAnnotator(false)); // pipeline.addAnnotator(new NumberAnnotator(false)); // pipeline.addAnnotator(new QuantifiableEntityNormalizingAnnotator(false, false)); String timeAnnotator = props.getProperty("timeAnnotator", "sutime"); switch (timeAnnotator) { case "gutime": useGUTime = true; pipeline.addAnnotator(new GUTimeAnnotator("gutime", props)); break; case "heideltime": requiredDocDateFormat = "yyyy-MM-dd"; pipeline.addAnnotator(new HeidelTimeAnnotator("heideltime", props)); break; case "sutime": pipeline.addAnnotator(new TimeAnnotator("sutime", props)); break; default: throw new IllegalArgumentException("Unknown timeAnnotator: " + timeAnnotator); } return pipeline; }
AnnotationPipeline ap = new AnnotationPipeline(); final boolean verbose = false; ap.addAnnotator(new TokenizerAnnotator(verbose, "en")); ap.addAnnotator(new WordsToSentencesAnnotator(verbose));
/** * Tokenize, emulating the Penn Treebank */ public Annotator tokenizer(Properties properties) { return new TokenizerAnnotator(properties); }
private static AnnotationPipeline makeNumericPipeline() { AnnotationPipeline pipeline = new AnnotationPipeline(); pipeline.addAnnotator(new TokenizerAnnotator(false, "en")); pipeline.addAnnotator(new WordsToSentencesAnnotator(false)); pipeline.addAnnotator(new POSTaggerAnnotator(false)); pipeline.addAnnotator(new TimeAnnotator(true)); return pipeline; }
public static AnnotationPipeline getPipeline(Properties props, boolean tokenize) throws Exception { // useGUTime = Boolean.parseBoolean(props.getProperty("gutime", "false")); AnnotationPipeline pipeline = new AnnotationPipeline(); if (tokenize) { pipeline.addAnnotator(new TokenizerAnnotator(false, "en")); pipeline.addAnnotator(new WordsToSentencesAnnotator(false)); } pipeline.addAnnotator(new POSTaggerAnnotator(false)); // pipeline.addAnnotator(new NumberAnnotator(false)); // pipeline.addAnnotator(new QuantifiableEntityNormalizingAnnotator(false, false)); String timeAnnotator = props.getProperty("timeAnnotator", "sutime"); switch (timeAnnotator) { case "gutime": useGUTime = true; pipeline.addAnnotator(new GUTimeAnnotator("gutime", props)); break; case "heideltime": requiredDocDateFormat = "yyyy-MM-dd"; pipeline.addAnnotator(new HeidelTimeAnnotator("heideltime", props)); break; case "sutime": pipeline.addAnnotator(new TimeAnnotator("sutime", props)); break; default: throw new IllegalArgumentException("Unknown timeAnnotator: " + timeAnnotator); } return pipeline; }
AnnotationPipeline ap = new AnnotationPipeline(); final boolean verbose = false; ap.addAnnotator(new TokenizerAnnotator(verbose, "en")); ap.addAnnotator(new WordsToSentencesAnnotator(verbose));