edu.stanford.nlp.international.spanish.process.AnCoraPronounDisambiguator Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of stanford-corenlp Show documentation
Show all versions of stanford-corenlp Show documentation
Stanford CoreNLP provides a set of natural language analysis tools which can take raw English language text input and give the base forms of words, their parts of speech, whether they are names of companies, people, etc., normalize dates, times, and numeric quantities, mark up the structure of sentences in terms of phrases and word dependencies, and indicate which noun phrases refer to the same entities. It provides the foundational building blocks for higher level text understanding applications.
package edu.stanford.nlp.international.spanish.process;
import edu.stanford.nlp.international.spanish.SpanishVerbStripper;
import edu.stanford.nlp.util.logging.Redwood;
import edu.stanford.nlp.util.Pair;
import java.util.Arrays;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
/**
* A utility for preprocessing the AnCora Spanish corpus.
*
* Attempts to disambiguate Spanish personal pronouns which have
* multiple senses:
*
* me, te, se, nos, os
*
* Each of these can be used as 1) an indirect object pronoun or as
* 2) a reflexive pronoun. (me, te, nos, and os can
* also be used as direct object pronouns.)
*
* For the purposes of corpus preprocessing, all we need is to
* distinguish between the object- and reflexive-pronoun cases.
*
* Disambiguation is done first by (dictionary-powered) heuristics, and
* then by brute force. The brute-force decisions are manual tags for
* verbs with clitic pronouns which appear in the AnCora corpus.
*
* @author Jon Gauthier
* @see edu.stanford.nlp.trees.international.spanish.SpanishTreeNormalizer
*/
public class AnCoraPronounDisambiguator {
/** A logger for this class */
private static Redwood.RedwoodChannels log = Redwood.channels(AnCoraPronounDisambiguator.class);
public static enum PersonalPronounType {OBJECT, REFLEXIVE, UNKNOWN}
private static final Set ambiguousPersonalPronouns = new HashSet<>(Arrays.asList(
"me", "te", "se", "nos", "os"
));
/**
* The following verbs always use ambiguous pronouns in a reflexive
* sense in the corpus.
*/
private static final Set alwaysReflexiveVerbs = new HashSet<>(Arrays.asList(
"acercar",
"acostumbrar",
"adaptar",
"afeitar",
"agarrar",
"ahincar",
"alegrar",
"Anticipar",
"aplicar",
"aprobar",
"aprovechar",
"asegurar",
"Atreve",
"bajar",
"beneficiar",
"callar",
"casar",
"cobrar",
"colocar",
"comer",
"comportar",
"comprar",
"concentrar",
"cuidar",
"deber",
"decidir",
"defender",
"desplazar",
"detectar",
"divirtiendo",
"echar",
"encontrar",
"enfrentar",
"entender",
"enterar",
"entrometer",
"equivocar",
"escapar",
"esconder",
"esforzando",
"establecer",
"felicitar",
"fija",
"Fija",
"ganar",
"guarda",
"guardar",
"Habituar",
"hacer",
"imagina",
"imaginar",
"iniciar",
"inscribir",
"ir",
"jode",
"jugar",
"Levantar",
"Manifestar",
"mantener",
"marchar",
"meter",
"Negar",
"obsesionar",
"Olvida",
"Olvidar",
"olvidar",
"oponer",
"Para",
"pasar",
"plantear",
"poner",
"pudra",
"queda",
"quedar",
"querer",
"quita",
"reciclar",
"reconoce",
"reconstruir",
"recordar",
"recuperar",
"reencontrar",
"referir",
"registrar",
"reincorporar",
"rendir",
"reservar",
"retirar",
"reunir",
"sentar",
"sentir",
"someter",
"subir",
"tirando",
"toma",
"tomar",
"tomen",
"Une",
"unir",
"Ve",
"vestir"
));
/**
* The following verbs always use ambiguous clitic pronouns in an
* object sense **in the corpora supported.**
*
* This does not imply that the below verbs are only ever non-reflexive!
* This list may need to be revised in order to produce correct gold trees
* on new datasets.
*/
private static final Set neverReflexiveVerbs = new HashSet<>(Arrays.asList(
"abrir",
"aguar",
"anunciar",
"arrebatando",
"arruinar",
"clasificar",
"compensar",
"compra",
"comprar",
"concretar",
"contar",
"crea",
"crear",
"Cuente",
"Decir",
"decir",
"deja",
"digan",
"devolver",
"devuelve",
"dirigiendo",
"distraer",
"enfrascar",
"exigiendo",
"exigir",
"haz",
"ignorar",
"impedir",
"insultar",
"juzgar",
"llamar",
"llevando",
"llevar",
"manda",
"mirar",
"Miren",
"multar",
"negar",
"ocultando",
"pagar",
"patear",
"pedir",
"permitir",
"pidiendo",
"preguntar",
"prevenir",
"quitar",
"razona",
"resultar",
"saca",
"sacar",
"saludar",
"seguir",
"servir",
"situar",
"suceder",
"tener",
"tutear",
"utilizar",
"vender",
"ver",
"visitar"
));
/**
* Brute-force: based on clauses which we recognize from AnCora,
* dictate the type of pronoun being used
*
* Map from pair (verb, containing clause) to personal pronoun type
*/
@SuppressWarnings("unchecked")
private static final Map, PersonalPronounType> bruteForceDecisions =
new HashMap<>();
static {
bruteForceDecisions.put(
new Pair<>("contar", "No contarte mi vida nunca más"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("Creer", "Creerselo todo"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("creer", "creérselo todo ..."), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("creer", "creerte"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("Dar", "Darte de alta ahi"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("da", "A mi dame billetes uno al lado del otro que es la forma mas líquida que uno pueda estar"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("da", "danos UNA razon UNA"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("da", "y ... dame una razon por la que hubiera matado o se hubiera comido a el compañero ?"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("dar", "darme cuenta"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("dar", "darme la enhorabuena"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("dar", "darnos cuenta"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("dar", "darselo a la doña"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("dar", "darte cuenta"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("dar", "darte de alta"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("dar", "darte vuelta en cuestiones que no tienen nada que ver con lo que comenzaste diciendo"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("dar", "podría darnos"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("dar", "puede darnos"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("decir", "suele decirnos"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("decir", "suelo decírmelo"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("dejar", "debería dejarnos faenar"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("dejar", "dejarme un intermitente encendido"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("dejar", ": dejarnos un país tan limpio en su gobierno como el cielo claro después de las tormentas mediterráneas , que inundan nuestras obras públicas sin encontrar nunca ni un solo responsable político de tanta mala gestión , ya sea la plaza de Cerdà socialista o los incendios forestales de la Generalitat"),
PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("dejar", "podemos dejarnos adormecer"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("engañar", "engañarnos"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("estira", "=LRB= al menos estirate a los japoneses HDP !!! =RRB="),
PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("explica", "explicame como hago"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("explicar", "deberá explicarnos"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("liar", "liarme a tiros"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("librar", "librarme de el mismo para siempre"),
PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("llevar", "llevarnos a una trampa en esta elección"),
PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("manifestar", "manifestarme su solidaridad"),
PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("manifestar", "manifestarnos sobre las circunstancias que mantienen en vilo la vida y obra de los colombianos"),
PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("mirando", "estábamos mirándonos"), PersonalPronounType.REFLEXIVE);
bruteForceDecisions.put(
new Pair<>("poner", "ponerme en ascuas"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("servir", "servirme de guía"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("volver", "debe volvernos"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("volver", "deja de volverme"), PersonalPronounType.OBJECT);
bruteForceDecisions.put(
new Pair<>("volver", "volvernos"), PersonalPronounType.REFLEXIVE);
}
/**
* Determine if the given pronoun can have multiple senses.
*/
public static boolean isAmbiguous(String pronoun) {
return ambiguousPersonalPronouns.contains(pronoun);
}
/**
* Determine whether the given clitic pronoun is an indirect object
* pronoun or a reflexive pronoun.
*
* This method is only defined when the pronoun is one of
*
* me, te, se, nos, os
*
* i.e., those in which the meaning is actually ambiguous.
*
* @param strippedVerb Stripped verb as returned by
* {@link edu.stanford.nlp.international.spanish.SpanishVerbStripper#separatePronouns(String)}.
* @param pronounIdx The index of the pronoun within
* {@code strippedVerb.getPronouns()} which should be
* disambiguated.
* @param clauseYield A string representing the yield of the
* clause which contains the given verb
* @throws java.lang.IllegalArgumentException If the given pronoun is
* not ambiguous, or its disambiguation is not supported.
*/
public static PersonalPronounType disambiguatePersonalPronoun(SpanishVerbStripper.StrippedVerb strippedVerb,
int pronounIdx, String clauseYield) {
List pronouns = strippedVerb.getPronouns();
String pronoun = pronouns.get(pronounIdx).toLowerCase();
if (!ambiguousPersonalPronouns.contains(pronoun))
throw new IllegalArgumentException("We don't support disambiguating pronoun '" + pronoun + "'");
if (pronouns.size() == 1 && pronoun.equalsIgnoreCase("se"))
return PersonalPronounType.REFLEXIVE;
String verb = strippedVerb.getStem();
if (alwaysReflexiveVerbs.contains(verb))
return PersonalPronounType.REFLEXIVE;
else if (neverReflexiveVerbs.contains(verb))
return PersonalPronounType.OBJECT;
Pair bruteForceKey = new Pair<>(verb, clauseYield);
if (bruteForceDecisions.containsKey(bruteForceKey))
return bruteForceDecisions.get(bruteForceKey);
// Log this instance where a clitic pronoun could not be disambiguated.
log.info("Failed to disambiguate: " + verb
+ "\nContaining clause:\t" + clauseYield + "\n");
return PersonalPronounType.UNKNOWN;
}
}