edu.stanford.nlp.quoteattribution

Class QuoteAttributionUtils

java.lang.Object

edu.stanford.nlp.quoteattribution.QuoteAttributionUtils

public class QuoteAttributionUtils
extends java.lang.Object

Various utility functions for Quote Attribution processing.

Author:

Grace Muzny, Michael Fang

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	QuoteAttributionUtils.EnhancedSentenceAnnotation

Constructor Summary

Constructors

Constructor and Description
QuoteAttributionUtils()

Method Summary

Modifier and Type	Method and Description
static void	addEnhancedSentences(Annotation doc, DependencyParser parser)
static void	annotateForDependencyParse(Annotation doc, DependencyParser parser)
static int	getParagraphBeginNumber(CoreMap quote)
static int	getParagraphEndNumber(CoreMap quote)
static int	getParagraphRank(Annotation doc, CoreMap quote)
static int	getQuoteParagraphIndex(Annotation doc, CoreMap quote)
static Pair<java.lang.Integer,java.lang.Integer>	getRemainderInSentence(Annotation doc, CoreMap quote) Take a context for a quote using the following preferences in order: (i) prior words in same sentence, if at least two, (ii) following words in same sentence, if at least two, (iii) pPrevious sentence, if any, (iv) next sentence.
static java.util.List<CoreMap>	getSentsForQuoteParagraphs(Annotation doc, CoreMap quote)
static java.util.List<CoreMap>	getSentsInParagraph(Annotation doc, int paragraph)
static Pair<java.lang.Boolean,Pair<java.lang.Integer,java.lang.Integer>>	getTokenRangeFollowingQuote(Annotation doc, CoreMap quote) Gets range of tokens that are in the same sentence as the beginning of the quote that follow it, if they exist, or the next sentence, if it is in the same paragraph.
static Pair<java.lang.Boolean,Pair<java.lang.Integer,java.lang.Integer>>	getTokenRangePrecedingQuote(Annotation doc, CoreMap quote) Gets range of tokens that are in the same sentence as the beginning of the quote that precede it, if they exist, or the previous sentence, if it is in the same paragraph.
static boolean	isPronominal(java.lang.String potentialPronoun)
protected static java.util.Map<java.lang.Integer,java.lang.String>	mapBammanToCharacterMap(java.util.Map<java.lang.Integer,java.util.List<CoreLabel>> BammanTokens, java.util.Map<java.lang.String,java.util.List<Person>> characterMap)
static boolean	rangeContains(Pair<java.lang.Integer,java.lang.Integer> r1, Pair<java.lang.Integer,java.lang.Integer> r2)
static java.util.Set<java.lang.String>	readAnimacyList(java.lang.String filename)
static java.util.ArrayList<Person>	readCharacterList(java.lang.String filename)
static java.util.Set<java.lang.String>	readFamilyRelations(java.lang.String filename)
static java.util.Map<java.lang.String,Person.Gender>	readGenderedNounList(java.lang.String filename)
static java.util.Map<java.lang.String,java.util.List<Person>>	readPersonMap(java.util.List<Person> personList)
static java.util.Map<java.lang.String,java.util.List<Person>>	readPersonMap(java.lang.String fileName)
static java.util.Map<java.lang.Integer,java.lang.String>	setupCoref(java.lang.String bammanFile, java.util.Map<java.lang.String,java.util.List<Person>> characterMap, Annotation doc) Create a map of coreference to canonical entities for pronouns.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

QuoteAttributionUtils

public QuoteAttributionUtils()

Method Detail

getRemainderInSentence

public static Pair<java.lang.Integer,java.lang.Integer> getRemainderInSentence(Annotation doc,
                                                                               CoreMap quote)

Take a context for a quote using the following preferences in order: (i) prior words in same sentence, if at least two, (ii) following words in same sentence, if at least two, (iii) pPrevious sentence, if any, (iv) next sentence. This still isn't perfect, when there is material in a sentence before and after the quote, the speech verb may be afterwards.

Parameters:

doc - The document we are analyzing

quote - The particular quote that we are finding a sentence remainder for in which to find speech verb

Returns:

A document token index range for the sentence or part sentence.

getQuoteParagraphIndex

public static int getQuoteParagraphIndex(Annotation doc,
                                         CoreMap quote)

addEnhancedSentences

public static void addEnhancedSentences(Annotation doc,
                                        DependencyParser parser)

getTokenRangePrecedingQuote

public static Pair<java.lang.Boolean,Pair<java.lang.Integer,java.lang.Integer>> getTokenRangePrecedingQuote(Annotation doc,
                                                                                                            CoreMap quote)

Gets range of tokens that are in the same sentence as the beginning of the quote that precede it, if they exist, or the previous sentence, if it is in the same paragraph. Return which you've used in extra argument. Also, ensure that the difference is at least two tokens.

Returns:

A Pair of whether same sentence followed by a Pair that is a token range in the document.

getTokenRangeFollowingQuote

public static Pair<java.lang.Boolean,Pair<java.lang.Integer,java.lang.Integer>> getTokenRangeFollowingQuote(Annotation doc,
                                                                                                            CoreMap quote)

Gets range of tokens that are in the same sentence as the beginning of the quote that follow it, if they exist, or the next sentence, if it is in the same paragraph. Return which you've used in extra argument. Also, ensure that the difference is at least two tokens.

Returns:

A Pair of whether same sentence followed by a Pair that is a token range in the document.

annotateForDependencyParse

public static void annotateForDependencyParse(Annotation doc,
                                              DependencyParser parser)

getParagraphRank

public static int getParagraphRank(Annotation doc,
                                   CoreMap quote)

getParagraphBeginNumber

public static int getParagraphBeginNumber(CoreMap quote)

getParagraphEndNumber

public static int getParagraphEndNumber(CoreMap quote)

getSentsInParagraph

public static java.util.List<CoreMap> getSentsInParagraph(Annotation doc,
                                                          int paragraph)

getSentsForQuoteParagraphs

public static java.util.List<CoreMap> getSentsForQuoteParagraphs(Annotation doc,
                                                                 CoreMap quote)

readGenderedNounList

public static java.util.Map<java.lang.String,Person.Gender> readGenderedNounList(java.lang.String filename)

readFamilyRelations

public static java.util.Set<java.lang.String> readFamilyRelations(java.lang.String filename)

readAnimacyList

public static java.util.Set<java.lang.String> readAnimacyList(java.lang.String filename)

readPersonMap

public static java.util.Map<java.lang.String,java.util.List<Person>> readPersonMap(java.util.List<Person> personList)

readPersonMap

public static java.util.Map<java.lang.String,java.util.List<Person>> readPersonMap(java.lang.String fileName)

readCharacterList

public static java.util.ArrayList<Person> readCharacterList(java.lang.String filename)

isPronominal

public static boolean isPronominal(java.lang.String potentialPronoun)

setupCoref

public static java.util.Map<java.lang.Integer,java.lang.String> setupCoref(java.lang.String bammanFile,
                                                                           java.util.Map<java.lang.String,java.util.List<Person>> characterMap,
                                                                           Annotation doc)

Create a map of coreference to canonical entities for pronouns. (todo: It'd be nice to also expand this to common nouns if they are coreferent to things!)

Parameters:

bammanFile - A file giving coreferences derived from analysis of David Bamman's bookNLP. This may be null, and then CoreNLP coref is used

characterMap - A mapping of names to people. This is only used for Bamman coreference, otherwise, it can be null. Or maybe not, maybe it is set up anyway from entity mentions....

doc - The CoreNLP document

Returns:

A Map from the pronoun word charOffsetBegin index (change to head's index!) to a String form of the canonical entity

mapBammanToCharacterMap

protected static java.util.Map<java.lang.Integer,java.lang.String> mapBammanToCharacterMap(java.util.Map<java.lang.Integer,java.util.List<CoreLabel>> BammanTokens,
                                                                                           java.util.Map<java.lang.String,java.util.List<Person>> characterMap)

rangeContains

public static boolean rangeContains(Pair<java.lang.Integer,java.lang.Integer> r1,
                                    Pair<java.lang.Integer,java.lang.Integer> r2)