public ModCollinsHeadFinder(AbstractTreebankLanguagePack tlp) : base(tlp, tlp.PunctuationTags())
{
// avoid punctuation as head in readonly default rule
NonTerminalInfo = new Dictionary<string, string[][]>();
// This version from Collins' diss (1999: 236-238)
// NNS, NN is actually sensible (money, etc.)!
// QP early isn't; should prefer JJR NN RB
// remove ADVP; it just shouldn't be there.
// if two JJ, should take right one (e.g. South Korean)
// NonTerminalInfo.Add(CoordinationTransformer.Adjective, new string[][]{{Left, PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.DollarSign, "QP"}, {Right, PartsOfSpeech.Adjective}, {Left, PartsOfSpeech.VerbPastParticiple, PartsOfSpeech.VerbGerundOrPresentParticiple, CoordinationTransformer.Adjective, "JJP", PartsOfSpeech.AdjectiveComparative, CoordinationTransformer.Noun, PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.Determiner, PartsOfSpeech.ForeignWord, PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative, "SBAR", PartsOfSpeech.Adverb}});
NonTerminalInfo.Add(CoordinationTransformer.Adjective,
new string[][]
{
new string[] {Left, PartsOfSpeech.DollarSign},
new string[]
{
RightDis, PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.Adjective,
QP, PartsOfSpeech.VerbPastParticiple, PartsOfSpeech.VerbGerundOrPresentParticiple
},
new string[] {Left, CoordinationTransformer.Adjective},
new string[]
{
RightDis, JJP, PartsOfSpeech.AdjectiveComparative, PartsOfSpeech.AdjectiveSuperlative,
PartsOfSpeech.Determiner, PartsOfSpeech.Adverb, PartsOfSpeech.AdverbComparative,
PartsOfSpeech.CardinalNumber, PartsOfSpeech.PrepositionOrSubordinateConjunction,
PartsOfSpeech.VerbPastTense
},
new string[] {Left, ADVP, CoordinationTransformer.Noun}
});
NonTerminalInfo.Add(JJP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.DollarSign,
QP, PartsOfSpeech.Adjective, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.VerbGerundOrPresentParticiple, CoordinationTransformer.Adjective, JJP, PartsOfSpeech.AdjectiveComparative,
CoordinationTransformer.Noun, PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.Determiner, PartsOfSpeech.ForeignWord,
PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative, SBAR, PartsOfSpeech.Adverb
}
});
// JJP is introduced for NML-like adjective phrases in Vadas' treebank; Chris wishes he hadn't used JJP which should be a POS-tag.
// ADVP rule rewritten by Chris in Nov 2010 to be rightdis. This is right! JJ.* is often head and rightmost.
NonTerminalInfo.Add(ADVP, new string[][]
{
new string[] {Left, ADVP, PartsOfSpeech.PrepositionOrSubordinateConjunction},
new string[]
{
RightDis, PartsOfSpeech.Adverb, PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative,
PartsOfSpeech.Adjective, PartsOfSpeech.AdjectiveComparative, PartsOfSpeech.AdjectiveSuperlative
},
new string[]
{
RightDis, PartsOfSpeech.Particle, PartsOfSpeech.Determiner, PartsOfSpeech.NounSingularOrMass,
PartsOfSpeech.CardinalNumber, CoordinationTransformer.Noun, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.ProperNounSingular, PartsOfSpeech.CoordinatingConjunction, PartsOfSpeech.ForeignWord,
PartsOfSpeech.NounPlural, CoordinationTransformer.Adjective, NML
}
});
NonTerminalInfo.Add(CONJP,
new string[][]
{
new string[]
{
Right, PartsOfSpeech.CoordinatingConjunction, PartsOfSpeech.Adverb,
PartsOfSpeech.PrepositionOrSubordinateConjunction
}
});
NonTerminalInfo.Add(FRAG, new string[][] {new string[] {Right}}); // crap
NonTerminalInfo.Add(INTJ, new string[][] {new string[] {Left}});
NonTerminalInfo.Add(LST,
new string[][] {new string[] {Right, PartsOfSpeech.ListItemMarker, PartsOfSpeech.ColonSemiColon}});
// NML is head in: (NAC-LOC (NML San Antonio) (, ,) (NNP Texas))
// TODO: NNP should be head (rare cases, could be ignored):
// (NAC (NML New York) (NNP Court) (PP of Appeals))
// (NAC (NML Prudential Insurance) (NNP Co.) (PP Of America))
// Chris: This could maybe still do with more thought, but NAC is rare.
NonTerminalInfo.Add(NAC,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.NounPlural, NML,
PartsOfSpeech.ProperNounSingular, PartsOfSpeech.ProperNounPlural, CoordinationTransformer.Noun, NAC,
PartsOfSpeech.ExistentialThere, PartsOfSpeech.DollarSign, PartsOfSpeech.CardinalNumber, QP,
PartsOfSpeech.PersonalPronoun, PartsOfSpeech.VerbGerundOrPresentParticiple,
PartsOfSpeech.Adjective,
PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.AdjectiveComparative, CoordinationTransformer.Adjective, JJP,
PartsOfSpeech.ForeignWord
}
});
// Added JJ to PP head table, since it is a head in several cases, e.g.:
// (PP (JJ next) (PP to them))
// When you have both JJ and IN daughters, it is invariably "such as" -- not so clear which should be head, but leave as IN
// should prefer JJ? (PP (JJ such) (IN as) (NP (NN crocidolite))) Michel thinks we should make JJ a head of PP
// added SYM as used in new treebanks for symbols filling role of IN
// Changed PP search to left -- just what you want for conjunction (and consistent with SemanticHeadFinder)
NonTerminalInfo.Add(PP,
new string[][]
{
new string[]
{
Right, PartsOfSpeech.PrepositionOrSubordinateConjunction, PartsOfSpeech.To,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.Particle, PartsOfSpeech.ForeignWord, PartsOfSpeech.Adjective, PartsOfSpeech.Symbol
},
new string[] {Left, PP}
});
NonTerminalInfo.Add(PRN,
new string[][]
{
new string[]
{
Left, VP, CoordinationTransformer.Noun, PP, SQ, S, SINV, SBAR, CoordinationTransformer.Adjective, JJP, ADVP, INTJ, WHNP,
NAC,
PartsOfSpeech.VerbNon3rdPersSingPresent, PartsOfSpeech.Adjective,
PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.ProperNounSingular
}
});
NonTerminalInfo.Add(PRT, new string[][] {new string[] {Right, PartsOfSpeech.Particle}});
// add '#' for pounds!!
NonTerminalInfo.Add(QP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.DollarSign, PartsOfSpeech.PrepositionOrSubordinateConjunction,
PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.Adjective,
PartsOfSpeech.CardinalNumber, PartsOfSpeech.Predeterminer, PartsOfSpeech.Determiner,
PartsOfSpeech.Adverb, NCD, QP, PartsOfSpeech.AdjectiveComparative,
PartsOfSpeech.AdjectiveSuperlative
}
});
// reduced relative clause can be any predicate VP, ADJP, NP, PP.
// For choosing between NP and PP, really need to know which one is temporal and to choose the other.
// It's not clear ADVP needs to be in the list at all (delete?).
NonTerminalInfo.Add(RRC,
new string[][] { new string[] { Left, RRC }, new string[] { Right, VP, CoordinationTransformer.Adjective, JJP, CoordinationTransformer.Noun, PP, ADVP } });
// delete IN -- go for main part of sentence; add FRAG
NonTerminalInfo.Add(S,
new string[][] { new string[] { Left, PartsOfSpeech.To, VP, S, FRAG, SBAR, CoordinationTransformer.Adjective, JJP, UCP, CoordinationTransformer.Noun } });
NonTerminalInfo.Add(SBAR,
new string[][]
{
new string[]
{
Left, WHNP, WHPP, WHADVP, WHADJP, PartsOfSpeech.PrepositionOrSubordinateConjunction,
PartsOfSpeech.Determiner, S, SQ, SINV, SBAR, FRAG
}
});
NonTerminalInfo.Add(SBARQ,
new string[][] {new string[] {Left, SQ, S, SINV, SBARQ, FRAG, SBAR}});
// cdm: if you have 2 VP under an SINV, you should really take the 2nd as syntactic head, because the first is a topicalized VP complement of the second, but for now I didn't change this, since it didn't help parsing. (If it were changed, it'd need to be also changed to the opposite in SemanticHeadFinder.)
NonTerminalInfo.Add(SINV,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbPastTense,
PartsOfSpeech.VerbNon3rdPersSingPresent, PartsOfSpeech.VerbBaseForm, PartsOfSpeech.Modal,
PartsOfSpeech.VerbPastParticiple, VP, S, SINV, CoordinationTransformer.Adjective, JJP, CoordinationTransformer.Noun
}
});
NonTerminalInfo.Add(SQ,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbPastTense,
PartsOfSpeech.VerbNon3rdPersSingPresent, PartsOfSpeech.VerbBaseForm, PartsOfSpeech.Modal, AUX,
AUXG, VP, SQ
}
});
// TODO: Should maybe put S before SQ for tag questions. Check.
NonTerminalInfo.Add(UCP, new string[][] {new string[] {Right}});
// below is weird!! Make 2 lists, one for good and one for bad heads??
// VP: added AUX and AUXG to work with Charniak tags
NonTerminalInfo.Add(VP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.To, PartsOfSpeech.VerbPastTense, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.Modal, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbBaseForm,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbNon3rdPersSingPresent, VP,
AUX, AUXG, CoordinationTransformer.Adjective, JJP,
PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.NounPlural, PartsOfSpeech.Adjective, CoordinationTransformer.Noun,
PartsOfSpeech.ProperNounSingular
}
});
NonTerminalInfo.Add(WHADJP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.WhAdverb, WHADVP, PartsOfSpeech.Adverb, PartsOfSpeech.Adjective, CoordinationTransformer.Adjective,
JJP, PartsOfSpeech.AdjectiveComparative
}
});
NonTerminalInfo.Add(WHADVP, new string[][] {new string[] {Right, PartsOfSpeech.WhAdverb, WHADVP}});
NonTerminalInfo.Add(WHNP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.WhDeterminer, PartsOfSpeech.WhPronoun, PartsOfSpeech.PossessiveWhPronoun,
WHADJP, WHPP, WHNP
}
});
NonTerminalInfo.Add(WHPP,
new string[][]
{
new string[]
{
Right, PartsOfSpeech.PrepositionOrSubordinateConjunction, PartsOfSpeech.To,
PartsOfSpeech.ForeignWord
}
});
NonTerminalInfo.Add(X,
new string[][] { new string[] { Right, S, VP, CoordinationTransformer.Adjective, JJP, CoordinationTransformer.Noun, SBAR, PP, X } });
NonTerminalInfo.Add(CoordinationTransformer.Noun,
new string[][]
{
new string[]
{
RightDis, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.ProperNounSingular,
PartsOfSpeech.ProperNounPlural, PartsOfSpeech.NounPlural, NML, NX,
PartsOfSpeech.PossessiveEnding, PartsOfSpeech.AdjectiveComparative
},
new string[] {Left, CoordinationTransformer.Noun, PartsOfSpeech.PersonalPronoun},
new string[] {RightDis, PartsOfSpeech.DollarSign, CoordinationTransformer.Adjective, JJP, PRN, PartsOfSpeech.ForeignWord},
new string[] {Right, PartsOfSpeech.CardinalNumber},
new string[]
{
RightDis, PartsOfSpeech.Adjective, PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.Adverb, QP,
PartsOfSpeech.Determiner, PartsOfSpeech.WhDeterminer, PartsOfSpeech.AdverbComparative, ADVP
}
});
NonTerminalInfo.Add(NX, NonTerminalInfo[CoordinationTransformer.Noun]);
// TODO: seems JJ should be head of NML in this case:
// (NP (NML (JJ former) (NML Red Sox) (JJ great)) (NNP Luis) (NNP Tiant)),
// (although JJ great is tagged wrong)
NonTerminalInfo.Add(NML, NonTerminalInfo[CoordinationTransformer.Noun]);
NonTerminalInfo.Add(POSSP, new string[][] {new string[] {Right, PartsOfSpeech.PossessiveEnding}});
/* HJT: Adding the following to deal with oddly formed data in (for example) the Brown corpus */
NonTerminalInfo.Add(ROOT, new string[][] {new string[] {Left, S, SQ, SINV, SBAR, FRAG}});
// Just to handle trees which have TOP instead of ROOT at the root
NonTerminalInfo.Add(TOP, NonTerminalInfo[ROOT]);
NonTerminalInfo.Add(TYPO, new string[][]
{
new string[]
{
Left, PartsOfSpeech.NounSingularOrMass, CoordinationTransformer.Noun, NML, PartsOfSpeech.ProperNounSingular,
PartsOfSpeech.ProperNounPlural, PartsOfSpeech.To,
PartsOfSpeech.VerbPastTense, PartsOfSpeech.VerbPastParticiple, PartsOfSpeech.Modal,
PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbBaseForm,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbNon3rdPersSingPresent, VP, CoordinationTransformer.Adjective,
JJP, FRAG
}
}); // for Brown (Roger)
NonTerminalInfo.Add(ADV, new string[][]
{
new string[]
{
Right, PartsOfSpeech.Adverb, PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative,
PartsOfSpeech.ForeignWord,
ADVP, PartsOfSpeech.To, PartsOfSpeech.CardinalNumber, PartsOfSpeech.AdjectiveComparative,
PartsOfSpeech.Adjective, PartsOfSpeech.PrepositionOrSubordinateConjunction, CoordinationTransformer.Noun, "NML",
PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.NounSingularOrMass
}
});
// SWBD
NonTerminalInfo.Add(EDITED, new string[][] {new string[] {Left}});
// crap rule for Switchboard (if don't delete EDITED nodes)
// in sw2756, a PartsOfSpeech.VerbBaseForm. (copy "VP" to handle this problem, though should really fix it on reading)
NonTerminalInfo.Add(PartsOfSpeech.VerbBaseForm,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.To, PartsOfSpeech.VerbPastTense, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.Modal, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbBaseForm,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbNon3rdPersSingPresent, VP,
AUX, AUXG, CoordinationTransformer.Adjective, JJP,
PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.NounPlural, PartsOfSpeech.Adjective, CoordinationTransformer.Noun,
PartsOfSpeech.ProperNounSingular
}
});
NonTerminalInfo.Add(META, new string[][] {new string[] {Left}});
// rule for OntoNotes, but maybe should just be deleted in TreeReader??
NonTerminalInfo.Add(XS, new string[][] {new string[] {Right, PartsOfSpeech.PrepositionOrSubordinateConjunction}});
// rule for new structure in QP, introduced by Stanford in QPTreeTransformer
// NonTerminalInfo.Add(null, new string[][] {{Left}}); // rule for OntoNotes from Michel, but it would be better to fix this in TreeReader or to use a default rule?
// todo: Uncomment this line if we always want to take the leftmost if no head rule is defined for the mother category.
// defaultRule = defaultLeftRule; // Don't exception, take leftmost if no rule defined for a certain parent category
}
public ModCollinsHeadFinder(AbstractTreebankLanguagePack tlp) : base(tlp, tlp.PunctuationTags())
{
// avoid punctuation as head in readonly default rule
NonTerminalInfo = new Dictionary <string, string[][]>();
// This version from Collins' diss (1999: 236-238)
// NNS, NN is actually sensible (money, etc.)!
// QP early isn't; should prefer JJR NN RB
// remove ADVP; it just shouldn't be there.
// if two JJ, should take right one (e.g. South Korean)
// NonTerminalInfo.Add(CoordinationTransformer.Adjective, new string[][]{{Left, PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.DollarSign, "QP"}, {Right, PartsOfSpeech.Adjective}, {Left, PartsOfSpeech.VerbPastParticiple, PartsOfSpeech.VerbGerundOrPresentParticiple, CoordinationTransformer.Adjective, "JJP", PartsOfSpeech.AdjectiveComparative, CoordinationTransformer.Noun, PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.Determiner, PartsOfSpeech.ForeignWord, PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative, "SBAR", PartsOfSpeech.Adverb}});
NonTerminalInfo.Add(CoordinationTransformer.Adjective,
new string[][]
{
new string[] { Left, PartsOfSpeech.DollarSign },
new string[]
{
RightDis, PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.Adjective,
QP, PartsOfSpeech.VerbPastParticiple, PartsOfSpeech.VerbGerundOrPresentParticiple
},
new string[] { Left, CoordinationTransformer.Adjective },
new string[]
{
RightDis, JJP, PartsOfSpeech.AdjectiveComparative, PartsOfSpeech.AdjectiveSuperlative,
PartsOfSpeech.Determiner, PartsOfSpeech.Adverb, PartsOfSpeech.AdverbComparative,
PartsOfSpeech.CardinalNumber, PartsOfSpeech.PrepositionOrSubordinateConjunction,
PartsOfSpeech.VerbPastTense
},
new string[] { Left, ADVP, CoordinationTransformer.Noun }
});
NonTerminalInfo.Add(JJP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.DollarSign,
QP, PartsOfSpeech.Adjective, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.VerbGerundOrPresentParticiple, CoordinationTransformer.Adjective, JJP, PartsOfSpeech.AdjectiveComparative,
CoordinationTransformer.Noun, PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.Determiner, PartsOfSpeech.ForeignWord,
PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative, SBAR, PartsOfSpeech.Adverb
}
});
// JJP is introduced for NML-like adjective phrases in Vadas' treebank; Chris wishes he hadn't used JJP which should be a POS-tag.
// ADVP rule rewritten by Chris in Nov 2010 to be rightdis. This is right! JJ.* is often head and rightmost.
NonTerminalInfo.Add(ADVP, new string[][]
{
new string[] { Left, ADVP, PartsOfSpeech.PrepositionOrSubordinateConjunction },
new string[]
{
RightDis, PartsOfSpeech.Adverb, PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative,
PartsOfSpeech.Adjective, PartsOfSpeech.AdjectiveComparative, PartsOfSpeech.AdjectiveSuperlative
},
new string[]
{
RightDis, PartsOfSpeech.Particle, PartsOfSpeech.Determiner, PartsOfSpeech.NounSingularOrMass,
PartsOfSpeech.CardinalNumber, CoordinationTransformer.Noun, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.ProperNounSingular, PartsOfSpeech.CoordinatingConjunction, PartsOfSpeech.ForeignWord,
PartsOfSpeech.NounPlural, CoordinationTransformer.Adjective, NML
}
});
NonTerminalInfo.Add(CONJP,
new string[][]
{
new string[]
{
Right, PartsOfSpeech.CoordinatingConjunction, PartsOfSpeech.Adverb,
PartsOfSpeech.PrepositionOrSubordinateConjunction
}
});
NonTerminalInfo.Add(FRAG, new string[][] { new string[] { Right } }); // crap
NonTerminalInfo.Add(INTJ, new string[][] { new string[] { Left } });
NonTerminalInfo.Add(LST,
new string[][] { new string[] { Right, PartsOfSpeech.ListItemMarker, PartsOfSpeech.ColonSemiColon } });
// NML is head in: (NAC-LOC (NML San Antonio) (, ,) (NNP Texas))
// TODO: NNP should be head (rare cases, could be ignored):
// (NAC (NML New York) (NNP Court) (PP of Appeals))
// (NAC (NML Prudential Insurance) (NNP Co.) (PP Of America))
// Chris: This could maybe still do with more thought, but NAC is rare.
NonTerminalInfo.Add(NAC,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.NounPlural, NML,
PartsOfSpeech.ProperNounSingular, PartsOfSpeech.ProperNounPlural, CoordinationTransformer.Noun, NAC,
PartsOfSpeech.ExistentialThere, PartsOfSpeech.DollarSign, PartsOfSpeech.CardinalNumber, QP,
PartsOfSpeech.PersonalPronoun, PartsOfSpeech.VerbGerundOrPresentParticiple,
PartsOfSpeech.Adjective,
PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.AdjectiveComparative, CoordinationTransformer.Adjective, JJP,
PartsOfSpeech.ForeignWord
}
});
// Added JJ to PP head table, since it is a head in several cases, e.g.:
// (PP (JJ next) (PP to them))
// When you have both JJ and IN daughters, it is invariably "such as" -- not so clear which should be head, but leave as IN
// should prefer JJ? (PP (JJ such) (IN as) (NP (NN crocidolite))) Michel thinks we should make JJ a head of PP
// added SYM as used in new treebanks for symbols filling role of IN
// Changed PP search to left -- just what you want for conjunction (and consistent with SemanticHeadFinder)
NonTerminalInfo.Add(PP,
new string[][]
{
new string[]
{
Right, PartsOfSpeech.PrepositionOrSubordinateConjunction, PartsOfSpeech.To,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.Particle, PartsOfSpeech.ForeignWord, PartsOfSpeech.Adjective, PartsOfSpeech.Symbol
},
new string[] { Left, PP }
});
NonTerminalInfo.Add(PRN,
new string[][]
{
new string[]
{
Left, VP, CoordinationTransformer.Noun, PP, SQ, S, SINV, SBAR, CoordinationTransformer.Adjective, JJP, ADVP, INTJ, WHNP,
NAC,
PartsOfSpeech.VerbNon3rdPersSingPresent, PartsOfSpeech.Adjective,
PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.ProperNounSingular
}
});
NonTerminalInfo.Add(PRT, new string[][] { new string[] { Right, PartsOfSpeech.Particle } });
// add '#' for pounds!!
NonTerminalInfo.Add(QP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.DollarSign, PartsOfSpeech.PrepositionOrSubordinateConjunction,
PartsOfSpeech.NounPlural, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.Adjective,
PartsOfSpeech.CardinalNumber, PartsOfSpeech.Predeterminer, PartsOfSpeech.Determiner,
PartsOfSpeech.Adverb, NCD, QP, PartsOfSpeech.AdjectiveComparative,
PartsOfSpeech.AdjectiveSuperlative
}
});
// reduced relative clause can be any predicate VP, ADJP, NP, PP.
// For choosing between NP and PP, really need to know which one is temporal and to choose the other.
// It's not clear ADVP needs to be in the list at all (delete?).
NonTerminalInfo.Add(RRC,
new string[][] { new string[] { Left, RRC }, new string[] { Right, VP, CoordinationTransformer.Adjective, JJP, CoordinationTransformer.Noun, PP, ADVP } });
// delete IN -- go for main part of sentence; add FRAG
NonTerminalInfo.Add(S,
new string[][] { new string[] { Left, PartsOfSpeech.To, VP, S, FRAG, SBAR, CoordinationTransformer.Adjective, JJP, UCP, CoordinationTransformer.Noun } });
NonTerminalInfo.Add(SBAR,
new string[][]
{
new string[]
{
Left, WHNP, WHPP, WHADVP, WHADJP, PartsOfSpeech.PrepositionOrSubordinateConjunction,
PartsOfSpeech.Determiner, S, SQ, SINV, SBAR, FRAG
}
});
NonTerminalInfo.Add(SBARQ,
new string[][] { new string[] { Left, SQ, S, SINV, SBARQ, FRAG, SBAR } });
// cdm: if you have 2 VP under an SINV, you should really take the 2nd as syntactic head, because the first is a topicalized VP complement of the second, but for now I didn't change this, since it didn't help parsing. (If it were changed, it'd need to be also changed to the opposite in SemanticHeadFinder.)
NonTerminalInfo.Add(SINV,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbPastTense,
PartsOfSpeech.VerbNon3rdPersSingPresent, PartsOfSpeech.VerbBaseForm, PartsOfSpeech.Modal,
PartsOfSpeech.VerbPastParticiple, VP, S, SINV, CoordinationTransformer.Adjective, JJP, CoordinationTransformer.Noun
}
});
NonTerminalInfo.Add(SQ,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbPastTense,
PartsOfSpeech.VerbNon3rdPersSingPresent, PartsOfSpeech.VerbBaseForm, PartsOfSpeech.Modal, AUX,
AUXG, VP, SQ
}
});
// TODO: Should maybe put S before SQ for tag questions. Check.
NonTerminalInfo.Add(UCP, new string[][] { new string[] { Right } });
// below is weird!! Make 2 lists, one for good and one for bad heads??
// VP: added AUX and AUXG to work with Charniak tags
NonTerminalInfo.Add(VP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.To, PartsOfSpeech.VerbPastTense, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.Modal, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbBaseForm,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbNon3rdPersSingPresent, VP,
AUX, AUXG, CoordinationTransformer.Adjective, JJP,
PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.NounPlural, PartsOfSpeech.Adjective, CoordinationTransformer.Noun,
PartsOfSpeech.ProperNounSingular
}
});
NonTerminalInfo.Add(WHADJP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.WhAdverb, WHADVP, PartsOfSpeech.Adverb, PartsOfSpeech.Adjective, CoordinationTransformer.Adjective,
JJP, PartsOfSpeech.AdjectiveComparative
}
});
NonTerminalInfo.Add(WHADVP, new string[][] { new string[] { Right, PartsOfSpeech.WhAdverb, WHADVP } });
NonTerminalInfo.Add(WHNP,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.WhDeterminer, PartsOfSpeech.WhPronoun, PartsOfSpeech.PossessiveWhPronoun,
WHADJP, WHPP, WHNP
}
});
NonTerminalInfo.Add(WHPP,
new string[][]
{
new string[]
{
Right, PartsOfSpeech.PrepositionOrSubordinateConjunction, PartsOfSpeech.To,
PartsOfSpeech.ForeignWord
}
});
NonTerminalInfo.Add(X,
new string[][] { new string[] { Right, S, VP, CoordinationTransformer.Adjective, JJP, CoordinationTransformer.Noun, SBAR, PP, X } });
NonTerminalInfo.Add(CoordinationTransformer.Noun,
new string[][]
{
new string[]
{
RightDis, PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.ProperNounSingular,
PartsOfSpeech.ProperNounPlural, PartsOfSpeech.NounPlural, NML, NX,
PartsOfSpeech.PossessiveEnding, PartsOfSpeech.AdjectiveComparative
},
new string[] { Left, CoordinationTransformer.Noun, PartsOfSpeech.PersonalPronoun },
new string[] { RightDis, PartsOfSpeech.DollarSign, CoordinationTransformer.Adjective, JJP, PRN, PartsOfSpeech.ForeignWord },
new string[] { Right, PartsOfSpeech.CardinalNumber },
new string[]
{
RightDis, PartsOfSpeech.Adjective, PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.Adverb, QP,
PartsOfSpeech.Determiner, PartsOfSpeech.WhDeterminer, PartsOfSpeech.AdverbComparative, ADVP
}
});
NonTerminalInfo.Add(NX, NonTerminalInfo[CoordinationTransformer.Noun]);
// TODO: seems JJ should be head of NML in this case:
// (NP (NML (JJ former) (NML Red Sox) (JJ great)) (NNP Luis) (NNP Tiant)),
// (although JJ great is tagged wrong)
NonTerminalInfo.Add(NML, NonTerminalInfo[CoordinationTransformer.Noun]);
NonTerminalInfo.Add(POSSP, new string[][] { new string[] { Right, PartsOfSpeech.PossessiveEnding } });
/* HJT: Adding the following to deal with oddly formed data in (for example) the Brown corpus */
NonTerminalInfo.Add(ROOT, new string[][] { new string[] { Left, S, SQ, SINV, SBAR, FRAG } });
// Just to handle trees which have TOP instead of ROOT at the root
NonTerminalInfo.Add(TOP, NonTerminalInfo[ROOT]);
NonTerminalInfo.Add(TYPO, new string[][]
{
new string[]
{
Left, PartsOfSpeech.NounSingularOrMass, CoordinationTransformer.Noun, NML, PartsOfSpeech.ProperNounSingular,
PartsOfSpeech.ProperNounPlural, PartsOfSpeech.To,
PartsOfSpeech.VerbPastTense, PartsOfSpeech.VerbPastParticiple, PartsOfSpeech.Modal,
PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbBaseForm,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbNon3rdPersSingPresent, VP, CoordinationTransformer.Adjective,
JJP, FRAG
}
}); // for Brown (Roger)
NonTerminalInfo.Add(ADV, new string[][]
{
new string[]
{
Right, PartsOfSpeech.Adverb, PartsOfSpeech.AdverbComparative, PartsOfSpeech.AdverbSuperlative,
PartsOfSpeech.ForeignWord,
ADVP, PartsOfSpeech.To, PartsOfSpeech.CardinalNumber, PartsOfSpeech.AdjectiveComparative,
PartsOfSpeech.Adjective, PartsOfSpeech.PrepositionOrSubordinateConjunction, CoordinationTransformer.Noun, "NML",
PartsOfSpeech.AdjectiveSuperlative, PartsOfSpeech.NounSingularOrMass
}
});
// SWBD
NonTerminalInfo.Add(EDITED, new string[][] { new string[] { Left } });
// crap rule for Switchboard (if don't delete EDITED nodes)
// in sw2756, a PartsOfSpeech.VerbBaseForm. (copy "VP" to handle this problem, though should really fix it on reading)
NonTerminalInfo.Add(PartsOfSpeech.VerbBaseForm,
new string[][]
{
new string[]
{
Left, PartsOfSpeech.To, PartsOfSpeech.VerbPastTense, PartsOfSpeech.VerbPastParticiple,
PartsOfSpeech.Modal, PartsOfSpeech.Verb3rdPersSingPresent, PartsOfSpeech.VerbBaseForm,
PartsOfSpeech.VerbGerundOrPresentParticiple, PartsOfSpeech.VerbNon3rdPersSingPresent, VP,
AUX, AUXG, CoordinationTransformer.Adjective, JJP,
PartsOfSpeech.NounSingularOrMass, PartsOfSpeech.NounPlural, PartsOfSpeech.Adjective, CoordinationTransformer.Noun,
PartsOfSpeech.ProperNounSingular
}
});
NonTerminalInfo.Add(META, new string[][] { new string[] { Left } });
// rule for OntoNotes, but maybe should just be deleted in TreeReader??
NonTerminalInfo.Add(XS, new string[][] { new string[] { Right, PartsOfSpeech.PrepositionOrSubordinateConjunction } });
// rule for new structure in QP, introduced by Stanford in QPTreeTransformer
// NonTerminalInfo.Add(null, new string[][] {{Left}}); // rule for OntoNotes from Michel, but it would be better to fix this in TreeReader or to use a default rule?
// todo: Uncomment this line if we always want to take the leftmost if no head rule is defined for the mother category.
// defaultRule = defaultLeftRule; // Don't exception, take leftmost if no rule defined for a certain parent category
}
public ModCollinsHeadFinder(AbstractTreebankLanguagePack tlp) : base(tlp, tlp.PunctuationTags())
{
// avoid punctuation as head in readonly default rule
nonTerminalInfo = new Dictionary <string, string[][]>();
// This version from Collins' diss (1999: 236-238)
// NNS, NN is actually sensible (money, etc.)!
// QP early isn't; should prefer JJR NN RB
// remove ADVP; it just shouldn't be there.
// if two JJ, should take right one (e.g. South Korean)
// nonTerminalInfo.Add("ADJP", new string[][]{{"left", "NNS", "NN", "$", "QP"}, {"right", "JJ"}, {"left", "VBN", "VBG", "ADJP", "JJP", "JJR", "NP", "JJS", "DT", "FW", "RBR", "RBS", "SBAR", "RB"}});
nonTerminalInfo.Add("ADJP",
new string[][]
{
new string[] { "left", "$" }, new string[] { "rightdis", "NNS", "NN", "JJ", "QP", "VBN", "VBG" },
new string[] { "left", "ADJP" },
new string[] { "rightdis", "JJP", "JJR", "JJS", "DT", "RB", "RBR", "CD", "IN", "VBD" },
new string[] { "left", "ADVP", "NP" }
});
nonTerminalInfo.Add("JJP",
new string[][]
{
new string[]
{
"left", "NNS", "NN", "$", "QP", "JJ", "VBN", "VBG", "ADJP", "JJP", "JJR", "NP", "JJS", "DT", "FW",
"RBR", "RBS", "SBAR", "RB"
}
});
// JJP is introduced for NML-like adjective phrases in Vadas' treebank; Chris wishes he hadn't used JJP which should be a POS-tag.
// ADVP rule rewritten by Chris in Nov 2010 to be rightdis. This is right! JJ.* is often head and rightmost.
nonTerminalInfo.Add("ADVP", new string[][]
{
new string[] { "left", "ADVP", "IN" },
new string[] { "rightdis", "RB", "RBR", "RBS", "JJ", "JJR", "JJS" },
new string[] { "rightdis", "RP", "DT", "NN", "CD", "NP", "VBN", "NNP", "CC", "FW", "NNS", "ADJP", "NML" }
});
nonTerminalInfo.Add("CONJP", new string[][] { new string[] { "right", "CC", "RB", "IN" } });
nonTerminalInfo.Add("FRAG", new string[][] { new string[] { "right" } }); // crap
nonTerminalInfo.Add("INTJ", new string[][] { new string[] { "left" } });
nonTerminalInfo.Add("LST", new string[][] { new string[] { "right", "LS", ":" } });
// NML is head in: (NAC-LOC (NML San Antonio) (, ,) (NNP Texas))
// TODO: NNP should be head (rare cases, could be ignored):
// (NAC (NML New York) (NNP Court) (PP of Appeals))
// (NAC (NML Prudential Insurance) (NNP Co.) (PP Of America))
// Chris: This could maybe still do with more thought, but NAC is rare.
nonTerminalInfo.Add("NAC",
new string[][]
{
new string[]
{
"left", "NN", "NNS", "NML", "NNP", "NNPS", "NP", "NAC", "EX", "$", "CD", "QP", "PRP", "VBG", "JJ",
"JJS", "JJR", "ADJP", "JJP", "FW"
}
});
// Added JJ to PP head table, since it is a head in several cases, e.g.:
// (PP (JJ next) (PP to them))
// When you have both JJ and IN daughters, it is invariably "such as" -- not so clear which should be head, but leave as IN
// should prefer JJ? (PP (JJ such) (IN as) (NP (NN crocidolite))) Michel thinks we should make JJ a head of PP
// added SYM as used in new treebanks for symbols filling role of IN
// Changed PP search to left -- just what you want for conjunction (and consistent with SemanticHeadFinder)
nonTerminalInfo.Add("PP",
new string[][]
{ new string[] { "right", "IN", "TO", "VBG", "VBN", "RP", "FW", "JJ", "SYM" }, new string[] { "left", "PP" } });
nonTerminalInfo.Add("PRN",
new string[][]
{
new string[]
{
"left", "VP", "NP", "PP", "SQ", "S", "SINV", "SBAR", "ADJP", "JJP", "ADVP", "INTJ", "WHNP", "NAC",
"VBP", "JJ", "NN", "NNP"
}
});
nonTerminalInfo.Add("PRT", new string[][] { new string[] { "right", "RP" } });
// add '#' for pounds!!
nonTerminalInfo.Add("QP",
new string[][]
{
new string[] { "left", "$", "IN", "NNS", "NN", "JJ", "CD", "PDT", "DT", "RB", "NCD", "QP", "JJR", "JJS" }
});
// reduced relative clause can be any predicate VP, ADJP, NP, PP.
// For choosing between NP and PP, really need to know which one is temporal and to choose the other.
// It's not clear ADVP needs to be in the list at all (delete?).
nonTerminalInfo.Add("RRC",
new string[][]
{ new string[] { "left", "RRC" }, new string[] { "right", "VP", "ADJP", "JJP", "NP", "PP", "ADVP" } });
// delete IN -- go for main part of sentence; add FRAG
nonTerminalInfo.Add("S",
new string[][] { new string[] { "left", "TO", "VP", "S", "FRAG", "SBAR", "ADJP", "JJP", "UCP", "NP" } });
nonTerminalInfo.Add("SBAR",
new string[][]
{
new string[]
{ "left", "WHNP", "WHPP", "WHADVP", "WHADJP", "IN", "DT", "S", "SQ", "SINV", "SBAR", "FRAG" }
});
nonTerminalInfo.Add("SBARQ",
new string[][] { new string[] { "left", "SQ", "S", "SINV", "SBARQ", "FRAG", "SBAR" } });
// cdm: if you have 2 VP under an SINV, you should really take the 2nd as syntactic head, because the first is a topicalized VP complement of the second, but for now I didn't change this, since it didn't help parsing. (If it were changed, it'd need to be also changed to the opposite in SemanticHeadFinder.)
nonTerminalInfo.Add("SINV",
new string[][]
{ new string[] { "left", "VBZ", "VBD", "VBP", "VB", "MD", "VBN", "VP", "S", "SINV", "ADJP", "JJP", "NP" } });
nonTerminalInfo.Add("SQ",
new string[][] { new string[] { "left", "VBZ", "VBD", "VBP", "VB", "MD", "AUX", "AUXG", "VP", "SQ" } });
// TODO: Should maybe put S before SQ for tag questions. Check.
nonTerminalInfo.Add("UCP", new string[][] { new string[] { "right" } });
// below is weird!! Make 2 lists, one for good and one for bad heads??
// VP: added AUX and AUXG to work with Charniak tags
nonTerminalInfo.Add("VP",
new string[][]
{
new string[]
{
"left", "TO", "VBD", "VBN", "MD", "VBZ", "VB", "VBG", "VBP", "VP", "AUX", "AUXG", "ADJP", "JJP",
"NN", "NNS", "JJ", "NP", "NNP"
}
});
nonTerminalInfo.Add("WHADJP",
new string[][] { new string[] { "left", "WRB", "WHADVP", "RB", "JJ", "ADJP", "JJP", "JJR" } });
nonTerminalInfo.Add("WHADVP", new string[][] { new string[] { "right", "WRB", "WHADVP" } });
nonTerminalInfo.Add("WHNP",
new string[][] { new string[] { "left", "WDT", "WP", "WP$", "WHADJP", "WHPP", "WHNP" } });
nonTerminalInfo.Add("WHPP", new string[][] { new string[] { "right", "IN", "TO", "FW" } });
nonTerminalInfo.Add("X",
new string[][] { new string[] { "right", "S", "VP", "ADJP", "JJP", "NP", "SBAR", "PP", "X" } });
nonTerminalInfo.Add("NP",
new string[][]
{
new string[] { "rightdis", "NN", "NNP", "NNPS", "NNS", "NML", "NX", "POS", "JJR" },
new string[] { "left", "NP", "PRP" }, new string[] { "rightdis", "$", "ADJP", "JJP", "PRN", "FW" },
new string[] { "right", "CD" },
new string[] { "rightdis", "JJ", "JJS", "RB", "QP", "DT", "WDT", "RBR", "ADVP" }
});
nonTerminalInfo.Add("NX", nonTerminalInfo["NP"]);
// TODO: seems JJ should be head of NML in this case:
// (NP (NML (JJ former) (NML Red Sox) (JJ great)) (NNP Luis) (NNP Tiant)),
// (although JJ great is tagged wrong)
nonTerminalInfo.Add("NML", nonTerminalInfo["NP"]);
nonTerminalInfo.Add("POSSP", new string[][] { new string[] { "right", "POS" } });
/* HJT: Adding the following to deal with oddly formed data in (for example) the Brown corpus */
nonTerminalInfo.Add("ROOT", new string[][] { new string[] { "left", "S", "SQ", "SINV", "SBAR", "FRAG" } });
// Just to handle trees which have TOP instead of ROOT at the root
nonTerminalInfo.Add("TOP", nonTerminalInfo["ROOT"]);
nonTerminalInfo.Add("TYPO", new string[][]
{
new string[]
{
"left", "NN", "NP", "NML", "NNP", "NNPS", "TO",
"VBD", "VBN", "MD", "VBZ", "VB", "VBG", "VBP", "VP", "ADJP", "JJP", "FRAG"
}
}); // for Brown (Roger)
nonTerminalInfo.Add("ADV", new string[][]
{
new string[]
{
"right", "RB", "RBR", "RBS", "FW",
"ADVP", "TO", "CD", "JJR", "JJ", "IN", "NP", "NML", "JJS", "NN"
}
});
// SWBD
nonTerminalInfo.Add("EDITED", new string[][] { new string[] { "left" } });
// crap rule for Switchboard (if don't delete EDITED nodes)
// in sw2756, a "VB". (copy "VP" to handle this problem, though should really fix it on reading)
nonTerminalInfo.Add("VB",
new string[][]
{
new string[]
{
"left", "TO", "VBD", "VBN", "MD", "VBZ", "VB", "VBG", "VBP", "VP", "AUX", "AUXG", "ADJP", "JJP",
"NN", "NNS", "JJ", "NP", "NNP"
}
});
nonTerminalInfo.Add("META", new string[][] { new string[] { "left" } });
// rule for OntoNotes, but maybe should just be deleted in TreeReader??
nonTerminalInfo.Add("XS", new string[][] { new string[] { "right", "IN" } });
// rule for new structure in QP, introduced by Stanford in QPTreeTransformer
// nonTerminalInfo.Add(null, new string[][] {{"left"}}); // rule for OntoNotes from Michel, but it would be better to fix this in TreeReader or to use a default rule?
// todo: Uncomment this line if we always want to take the leftmost if no head rule is defined for the mother category.
// defaultRule = defaultLeftRule; // Don't exception, take leftmost if no rule defined for a certain parent category
}