/**
* Lemmatise a phrase or word. If a phrase, only lemmatise the most RHS word.
*
* @param value
* @return
*/
public String normalise(String value) {
if (value.indexOf(" ") == -1
|| value.endsWith(" s")
|| value.endsWith(
"'s")) // if string is a single word, or it is in "XYZ's" form where the ' char has been
// removed
return lemmatizer.lemmatize(value, 1).trim();
String part1 = value.substring(0, value.lastIndexOf(" "));
String part2 = lemmatizer.lemmatize(value.substring(value.lastIndexOf(" ") + 1), 1);
return part1 + " " + part2.trim();
}
/**
* Outputs the settings for this configuration to the console, very useful for ensuring the
* configuration is set as desired prior to a training run
*/
public void log() {
System.out.println(
"Lemmatiser: " + (lemmatiser == null ? null : lemmatiser.getClass().getName()));
System.out.println("POSTagger: " + (posTagger == null ? null : posTagger.getClass().getName()));
System.out.println("Tokenizer: " + tokenizer.getClass().getName());
System.out.println("Tag format: " + tagFormat.name());
System.out.println(
"PostProcessor: " + (postProcessor == null ? null : postProcessor.getClass().getName()));
System.out.println("Using numeric normalization: " + useNumericNormalization);
System.out.println("CRF order is " + order);
System.out.println("Using feature induction: " + useFeatureInduction);
System.out.println("Text textDirection: " + textDirection);
}
/**
* @param value original word
* @param pos the part of speech of the last word
* @return the lemma of original word
*/
public String getLemma(String value, String pos) {
int POS = tagLookUp.get(pos);
if (POS == 0) return lemmatizer.lemmatize(value);
else return lemmatizer.lemmatize(value, POS);
}
