public override bool CanResolve(MentionContext mention)
{
string tag = mention.HeadTokenTag;
return(tag != null && PartsOfSpeech.IsPersonalOrPossessivePronoun(tag) &&
Linker.PluralThirdPersonPronounPattern.IsMatch(mention.HeadTokenText));
}
protected internal override bool IsOutOfRange(MentionContext mention, DiscourseEntity entity)
{
MentionContext cec = entity.LastExtent;
//System.err.println("MaxentSingularPronounresolve.outOfRange: ["+entity.getLastExtent().toText()+" ("+entity.getId()+")] ["+mention.toText()+" ("+mention.getId()+")] entity.sentenceNumber=("+entity.getLastExtent().getSentenceNumber()+")-mention.sentenceNumber=("+mention.getSentenceNumber()+") > "+numSentencesBack);
return(mention.SentenceNumber - cec.SentenceNumber > NumberSentencesBack);
}
protected internal override List<string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List<string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
features.AddRange(GetPronounMatchFeatures(mention, entity));
List<string> contexts = GetContextFeatures(mention);
MentionContext cec = entity.LastExtent;
if (mention.HeadTokenTag.StartsWith("PRP") && cec.HeadTokenTag.StartsWith("PRP"))
{
features.Add(mention.HeadTokenText + "," + cec.HeadTokenText);
}
else if (mention.HeadTokenText.StartsWith("NNP"))
{
for (int ci = 0, cl = contexts.Count; ci < cl; ci++)
{
features.Add(contexts[ci]);
}
features.Add(mention.NameType + "," + cec.HeadTokenText);
}
else
{
List<string> ccontexts = GetContextFeatures(cec);
for (int ci = 0, cl = ccontexts.Count; ci < cl; ci++)
{
features.Add(ccontexts[ci]);
}
features.Add(cec.NameType + "," + mention.HeadTokenText);
}
}
return (features);
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return(true);
}
string mentionGender = null;
foreach (MentionContext entityMention in entity.Mentions)
{
string tag = entityMention.HeadTokenTag;
if (tag != null && PartsOfSpeech.IsPersOrPossPronoun(tag) && Linker.SingularThirdPersonPronounPattern.IsMatch(mention.HeadTokenText))
{
if (mentionGender == null)
{
//lazy initilization
mentionGender = GetPronounGender(mention.HeadTokenText);
}
string entityGender = GetPronounGender(entityMention.HeadTokenText);
if (!entityGender.Equals("u") && !mentionGender.Equals(entityGender))
{
return(true);
}
}
}
return(false);
}
protected internal override bool IsExcluded(MentionContext context, DiscourseEntity discourseEntity)
{
MentionContext currentContext = discourseEntity.LastExtent;
//System.err.println("IsAResolver.excluded?: ec.span="+ec.getSpan()+" cec.span="+cec.getSpan()+" cec="+cec.toText()+" lastToken="+ec.getNextToken());
if (context.SentenceNumber != currentContext.SentenceNumber)
{
//System.err.println("IsAResolver.excluded: (true) not same sentence");
return(true);
}
//shallow parse appositives
//System.err.println("IsAResolver.excluded: ec="+ec.toText()+" "+ec.span+" cec="+cec.toText()+" "+cec.span);
if (currentContext.IndexSpan.End == context.IndexSpan.Start - 2)
{
return(false);
}
//full parse w/o trailing comma
if (currentContext.IndexSpan.End == context.IndexSpan.End)
{
//System.err.println("IsAResolver.excluded: (false) spans share end");
return(false);
}
//full parse w/ trailing comma or period
if (currentContext.IndexSpan.End <= context.IndexSpan.End + 2 && (context.NextToken != null && (context.NextToken.ToString().Equals(",") || context.NextToken.ToString().Equals("."))))
{
//System.err.println("IsAResolver.excluded: (false) spans end + punct");
return(false);
}
//System.err.println("IsAResolver.excluded: (true) default");
return(true);
}
protected internal override List <string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List <string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
features.AddRange(GetPronounMatchFeatures(mention, entity));
List <string> contexts = GetContextFeatures(mention);
MentionContext cec = entity.LastExtent;
if (mention.HeadTokenTag.StartsWith("PRP") && cec.HeadTokenTag.StartsWith("PRP"))
{
features.Add(mention.HeadTokenText + "," + cec.HeadTokenText);
}
else if (mention.HeadTokenText.StartsWith("NNP"))
{
for (int ci = 0, cl = contexts.Count; ci < cl; ci++)
{
features.Add(contexts[ci]);
}
features.Add(mention.NameType + "," + cec.HeadTokenText);
}
else
{
List <string> ccontexts = GetContextFeatures(cec);
for (int ci = 0, cl = ccontexts.Count; ci < cl; ci++)
{
features.Add(ccontexts[ci]);
}
features.Add(cec.NameType + "," + mention.HeadTokenText);
}
}
return(features);
}
public override bool CanResolve(MentionContext mention)
{
//System.err.println("MaxentSingularPronounResolver.canResolve: ec= ("+mention.id+") "+ mention.toText());
string tag = mention.HeadTokenTag;
return(tag != null && tag.StartsWith("PRP") && Linker.SingularThirdPersonPronounPattern.IsMatch(mention.HeadTokenText));
}
public override bool CanResolve(MentionContext mention)
{
string firstTok = mention.FirstTokenText.ToLower();
string firstTokTag = mention.FirstToken.SyntacticType;
bool rv = mention.HeadTokenTag.Equals("NNS") && !IsDefiniteArticle(firstTok, firstTokTag);
return rv;
}
protected internal override List <string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
var features = base.GetFeatures(mention, entity);
if (entity != null)
{
features.AddRange(GetPronounMatchFeatures(mention, entity));
var contexts = GetContextFeatures(mention);
var cec = entity.LastExtent;
if (PartsOfSpeech.IsPersOrPossPronoun(mention.HeadTokenTag) && PartsOfSpeech.IsPersOrPossPronoun(cec.HeadTokenTag))
{
features.Add(mention.HeadTokenText + "," + cec.HeadTokenText);
}
else if (PartsOfSpeech.IsProperNoun(mention.HeadTokenText))
{
for (int ci = 0, cl = contexts.Count; ci < cl; ci++)
{
features.Add(contexts[ci]);
}
features.Add(mention.NameType + "," + cec.HeadTokenText);
}
else
{
var ccontexts = GetContextFeatures(cec);
for (int ci = 0, cl = ccontexts.Count; ci < cl; ci++)
{
features.Add(ccontexts[ci]);
}
features.Add(cec.NameType + "," + mention.HeadTokenText);
}
}
return(features);
}
protected internal override bool IsExcluded(MentionContext context, DiscourseEntity discourseEntity)
{
MentionContext currentContext = discourseEntity.LastExtent;
if (context.SentenceNumber != currentContext.SentenceNumber)
{
return true;
}
//shallow parse appositives
if (currentContext.IndexSpan.End == context.IndexSpan.Start - 2)
{
return false;
}
//full parse w/o trailing comma
if (currentContext.IndexSpan.End == context.IndexSpan.End)
{
return false;
}
//full parse w/ trailing comma or period
if (currentContext.IndexSpan.End <= context.IndexSpan.End + 2 && (context.NextToken != null
&& (context.NextToken.ToString() == PartsOfSpeech.Comma || context.NextToken.ToString() == PartsOfSpeech.SentenceFinalPunctuation)))
{
return false;
}
return true;
}
public override bool CanResolve(MentionContext mention)
{
string firstTok = mention.FirstTokenText.ToLower();
string firstTokTag = mention.FirstToken.SyntacticType;
bool rv = mention.HeadTokenTag == PartsOfSpeech.NounPlural && !IsDefiniteArticle(firstTok, firstTokTag);
return rv;
}
protected internal override bool IsExcluded(MentionContext context, DiscourseEntity discourseEntity)
{
MentionContext currentContext = discourseEntity.LastExtent;
if (context.SentenceNumber != currentContext.SentenceNumber)
{
return(true);
}
//shallow parse appositives
if (currentContext.IndexSpan.End == context.IndexSpan.Start - 2)
{
return(false);
}
//full parse w/o trailing comma
if (currentContext.IndexSpan.End == context.IndexSpan.End)
{
return(false);
}
//full parse w/ trailing comma or period
if (currentContext.IndexSpan.End <= context.IndexSpan.End + 2 && (context.NextToken != null &&
(context.NextToken.ToString() == PartsOfSpeech.Comma || context.NextToken.ToString() == PartsOfSpeech.SentenceFinalPunctuation)))
{
return(false);
}
return(true);
}
protected internal override bool IsOutOfRange(MentionContext mention, DiscourseEntity entity)
{
MentionContext cec = entity.LastExtent;
//System.err.println("MaxentPluralPronounResolver.outOfRange: ["+ec.toText()+" ("+ec.id+")] ["+cec.toText()+" ("+cec.id+")] ec.sentenceNumber=("+ec.sentenceNumber+")-cec.sentenceNumber=("+cec.sentenceNumber+") > "+NUM_SENTS_BACK_PRONOUNS);
return(mention.SentenceNumber - cec.SentenceNumber > NUM_SENTS_BACK_PRONOUNS);
}
public override bool CanResolve(MentionContext mention)
{
string tag = mention.HeadTokenTag;
bool fpp = tag != null && tag.StartsWith("PRP") && Linker.SpeechPronounPattern.IsMatch(mention.HeadTokenText);
bool pn = tag != null && tag.StartsWith("NNP");
return (fpp || pn);
}
public override bool CanResolve(MentionContext mention)
{
var tag = mention.HeadTokenTag;
var fpp = tag != null && PartsOfSpeech.IsPersOrPossPronoun(tag) && Linker.SpeechPronounPattern.IsMatch(mention.HeadTokenText);
var pn = tag != null && PartsOfSpeech.IsProperNoun(tag);
return(fpp || pn);
}
public override bool CanResolve(MentionContext mention)
{
string tag = mention.HeadTokenTag;
bool fpp = tag != null && tag.StartsWith("PRP") && Linker.SpeechPronounPattern.IsMatch(mention.HeadTokenText);
bool pn = tag != null && tag.StartsWith("NNP");
return(fpp || pn);
}
public override bool CanResolve(MentionContext context)
{
if (PartsOfSpeech.IsNoun(context.HeadTokenTag))
{
return (context.PreviousToken != null && predicativePattern.IsMatch(context.PreviousToken.ToString()));
}
return false;
}
public override bool CanResolve(MentionContext context)
{
if (PartsOfSpeech.IsNoun(context.HeadTokenTag))
{
return(context.PreviousToken != null && predicativePattern.IsMatch(context.PreviousToken.ToString()));
}
return(false);
}
public override bool CanResolve(MentionContext mention)
{
string firstTok = mention.FirstTokenText.ToLower();
string firstTokTag = mention.FirstToken.SyntacticType;
bool rv = mention.HeadTokenTag.Equals("NNS") && !IsDefiniteArticle(firstTok, firstTokTag);
return(rv);
}
public override bool CanResolve(MentionContext context)
{
if (context.HeadTokenTag.StartsWith("NN"))
{
return(context.PreviousToken != null && predicativePattern.IsMatch(context.PreviousToken.ToString()));
}
return(false);
}
public override bool CanResolve(MentionContext mention)
{
Object[] mtokens = mention.Tokens;
string firstTok = mention.FirstTokenText.ToLower();
bool rv = mtokens.Length > 1 && !PartsOfSpeech.IsProperNoun(mention.HeadTokenTag) && IsDefiniteArticle(firstTok, mention.FirstTokenTag);
return (rv);
}
public override bool CanResolve(MentionContext mention)
{
string firstTok = mention.FirstTokenText.ToLower();
string firstTokTag = mention.FirstToken.SyntacticType;
bool rv = mention.HeadTokenTag == PartsOfSpeech.NounPlural && !IsDefiniteArticle(firstTok, firstTokTag);
return(rv);
}
public override bool CanResolve(MentionContext mention)
{
string tag = mention.HeadTokenTag;
bool fpp = tag != null && PartsOfSpeech.IsPersonalOrPossessivePronoun(tag) && Linker.SpeechPronounPattern.IsMatch(mention.HeadTokenText);
bool pn = tag != null && PartsOfSpeech.IsProperNoun(tag);
return(fpp || pn);
}
public override bool CanResolve(MentionContext context)
{
if (context.HeadTokenTag.StartsWith("NN"))
{
return (context.PreviousToken != null && predicativePattern.IsMatch(context.PreviousToken.ToString()));
}
return false;
}
public override bool CanResolve(MentionContext mention)
{
Object[] mtokens = mention.Tokens;
string firstTok = mention.FirstTokenText.ToLower();
bool rv = mtokens.Length > 1 && !PartsOfSpeech.IsProperNoun(mention.HeadTokenTag) && IsDefiniteArticle(firstTok, mention.FirstTokenTag);
return(rv);
}
protected internal override List<string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List<string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
features.AddRange(GetContextFeatures(mention));
features.AddRange(GetStringMatchFeatures(mention, entity));
}
return features;
}
protected internal override List <string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List <string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
features.AddRange(GetStringMatchFeatures(mention, entity));
features.AddRange(getAcronymFeatures(mention, entity));
}
return(features);
}
public override bool CanResolve(MentionContext mention)
{
System.Object[] mtokens = mention.Tokens;
string firstTok = mention.FirstTokenText.ToLower();
bool rv = mtokens.Length > 1 && !mention.HeadTokenTag.StartsWith("NNP") && IsDefiniteArticle(firstTok, mention.FirstTokenTag);
//if (rv) {
// System.err.println("defNp "+ec);
//}
return (rv);
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return true;
}
else
{
MentionContext cec = entity.LastExtent;
return (cec.HeadTokenTag != PartsOfSpeech.NounPlural || base.IsExcluded(mention, entity));
}
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return(true);
}
else
{
MentionContext cec = entity.LastExtent;
return(!cec.HeadTokenTag.Equals("NNS") || base.IsExcluded(mention, entity));
}
}
public override bool CanResolve(MentionContext mention)
{
System.Object[] mtokens = mention.Tokens;
string firstTok = mention.FirstTokenText.ToLower();
bool rv = mtokens.Length > 1 && !mention.HeadTokenTag.StartsWith("NNP") && IsDefiniteArticle(firstTok, mention.FirstTokenTag);
//if (rv) {
// System.err.println("defNp "+ec);
//}
return(rv);
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return(true);
}
else
{
MentionContext cec = entity.LastExtent;
return(cec.HeadTokenTag != PartsOfSpeech.NounPlural || base.IsExcluded(mention, entity));
}
}
protected internal override List <string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
//System.err.println("ProperNounResolver.getFeatures: "+mention.toText()+" -> "+entity);
List <string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
features.AddRange(GetStringMatchFeatures(mention, entity));
features.AddRange(getAcronymFeatures(mention, entity));
}
return(features);
}
//UPGRADE_NOTE: Access modifiers of method 'excluded' were changed to 'public'. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1204'"
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return(true);
}
var cec = entity.LastExtent;
if (!CanResolve(cec))
{
return(true);
}
if (PartsOfSpeech.IsProperNoun(mention.HeadTokenTag))
{
//mention is a propernoun
if (PartsOfSpeech.IsProperNoun(cec.HeadTokenTag))
{
return(true); // both NNP
}
else
{
if (entity.MentionCount > 1)
{
return(true);
}
return(!CanResolve(cec));
}
}
else if (PartsOfSpeech.IsPersOrPossPronoun(mention.HeadTokenTag))
{
// mention is a speech pronoun
// cec can be either a speech pronoun or a propernoun
if (PartsOfSpeech.IsProperNoun(cec.HeadTokenTag))
{
//exclude antecedents not in the same sentence when they are not pronoun
return(mention.SentenceNumber - cec.SentenceNumber != 0);
}
else if (PartsOfSpeech.IsPersOrPossPronoun(cec.HeadTokenTag))
{
return(false);
}
else
{
Console.Error.WriteLine("Unexpected candidate exluded: " + cec.ToText());
return(true);
}
}
else
{
Console.Error.WriteLine("Unexpected mention exluded: " + mention.ToText());
return(true);
}
}
protected internal override List <string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List <string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
//generate pronoun w/ referent features
MentionContext cec = entity.LastExtent;
//string gen = getPronounGender(pronoun);
features.AddRange(GetPronounMatchFeatures(mention, entity));
features.AddRange(GetContextFeatures(cec));
features.AddRange(GetDistanceFeatures(mention, entity));
features.Add(GetMentionCountFeature(entity));
/*
* //lexical features
* Set featureSet = new HashSet();
* for (Iterator ei = entity.getExtents(); ei.hasNext();) {
* MentionContext ec = (MentionContext) ei.next();
* List toks = ec.tokens;
* Parse tok;
* int headIndex = PTBHeadFinder.getInstance().getHeadIndex(toks);
* for (int ti = 0; ti < headIndex; ti++) {
* tok = (Parse) toks.get(ti);
* featureSet.add(gen + "mw=" + tok.ToString().toLowerCase());
* featureSet.add(gen + "mt=" + tok.getSyntacticType());
* }
* tok = (Parse) toks.get(headIndex);
* featureSet.add(gen + "hw=" + tok.ToString().toLowerCase());
* featureSet.add(gen + "ht=" + tok.getSyntacticType());
* //semantic features
* if (ec.neType != null) {
* featureSet.add(gen + "," + ec.neType);
* }
* else {
* for (Iterator si = ec.synsets.iterator(); si.hasNext();) {
* Integer synset = (Integer) si.next();
* featureSet.add(gen + "," + synset);
* }
* }
* }
* Iterator fset = featureSet.iterator();
* while (fset.hasNext()) {
* string f = (string) fset.next();
* features.add(f);
* }
*/
}
return(features);
}
protected internal override List<string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List<string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
//generate pronoun w/ referent features
MentionContext cec = entity.LastExtent;
//string gen = getPronounGender(pronoun);
features.AddRange(GetPronounMatchFeatures(mention, entity));
features.AddRange(GetContextFeatures(cec));
features.AddRange(GetDistanceFeatures(mention, entity));
features.Add(GetMentionCountFeature(entity));
/*
//lexical features
Set featureSet = new HashSet();
for (Iterator ei = entity.getExtents(); ei.hasNext();) {
MentionContext ec = (MentionContext) ei.next();
List toks = ec.tokens;
Parse tok;
int headIndex = PTBHeadFinder.getInstance().getHeadIndex(toks);
for (int ti = 0; ti < headIndex; ti++) {
tok = (Parse) toks.get(ti);
featureSet.add(gen + "mw=" + tok.toString().toLowerCase());
featureSet.add(gen + "mt=" + tok.getSyntacticType());
}
tok = (Parse) toks.get(headIndex);
featureSet.add(gen + "hw=" + tok.toString().toLowerCase());
featureSet.add(gen + "ht=" + tok.getSyntacticType());
//semantic features
if (ec.neType != null) {
featureSet.add(gen + "," + ec.neType);
}
else {
for (Iterator si = ec.synsets.iterator(); si.hasNext();) {
Integer synset = (Integer) si.next();
featureSet.add(gen + "," + synset);
}
}
}
Iterator fset = featureSet.iterator();
while (fset.hasNext()) {
string f = (string) fset.next();
features.add(f);
}
*/
}
return (features);
}
protected internal virtual List <string> getAcronymFeatures(MentionContext mention, DiscourseEntity entity)
{
MentionContext xec = getProperNounExtent(entity);
string ecStrip = StripNounPhrase(mention);
string xecStrip = StripNounPhrase(xec);
if (ecStrip != null && xecStrip != null)
{
if (isAcronym(ecStrip, xecStrip))
{
List <string> features = new List <string>(1);
features.Add("knownAcronym");
return(features);
}
}
return(new List <string>());
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return(true);
}
foreach (MentionContext xec in entity.Mentions)
{
if (PartsOfSpeech.IsProperNoun(xec.HeadTokenTag))
{
// || initialCaps.matcher(xec.headToken.ToString()).find()) {
return(false);
}
}
return(true);
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return(true);
}
foreach (MentionContext xec in entity.Mentions)
{
if (xec.HeadTokenTag.StartsWith("NNP"))
{
// || initialCaps.matcher(xec.headToken.toString()).find()) {
//System.err.println("MaxentProperNounResolver.exclude: kept "+xec.toText()+" with "+xec.headTag);
return(false);
}
}
return(true);
}
protected internal override List <string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List <string> features = base.GetFeatures(mention, entity);
//features.add("eid="+pc.id);
if (entity != null)
{
//generate pronoun w/ referent features
features.AddRange(GetPronounMatchFeatures(mention, entity));
MentionContext cec = entity.LastExtent;
features.AddRange(GetDistanceFeatures(mention, entity));
features.AddRange(GetContextFeatures(cec));
features.Add(GetMentionCountFeature(entity));
/*
* //lexical features
* Set featureSet = new HashSet();
* for (Iterator ei = entity.getExtents(); ei.hasNext();) {
* MentionContext ec = (MentionContext) ei.next();
* int headIndex = PTBHeadFinder.getInstance().getHeadIndex(ec.tokens);
* Parse tok = (Parse) ec.tokens.get(headIndex);
* featureSet.add("hw=" + tok.ToString().toLowerCase());
* if (ec.parse.isCoordinatedNounPhrase()) {
* featureSet.add("ht=CC");
* }
* else {
* featureSet.add("ht=" + tok.getSyntacticType());
* }
* if (ec.neType != null){
* featureSet.add("ne="+ec.neType);
* }
* }
* Iterator fset = featureSet.iterator();
* while (fset.hasNext()) {
* string f = (string) fset.next();
* features.add(f);
* }
*/
}
return(features);
}
protected internal override List<string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List<string> features = base.GetFeatures(mention, entity);
//features.add("eid="+pc.id);
if (entity != null)
{
//generate pronoun w/ referent features
features.AddRange(GetPronounMatchFeatures(mention, entity));
MentionContext cec = entity.LastExtent;
features.AddRange(GetDistanceFeatures(mention, entity));
features.AddRange(GetContextFeatures(cec));
features.Add(GetMentionCountFeature(entity));
/*
//lexical features
Set featureSet = new HashSet();
for (Iterator ei = entity.getExtents(); ei.hasNext();) {
MentionContext ec = (MentionContext) ei.next();
int headIndex = PTBHeadFinder.getInstance().getHeadIndex(ec.tokens);
Parse tok = (Parse) ec.tokens.get(headIndex);
featureSet.add("hw=" + tok.toString().toLowerCase());
if (ec.parse.isCoordinatedNounPhrase()) {
featureSet.add("ht=CC");
}
else {
featureSet.add("ht=" + tok.getSyntacticType());
}
if (ec.neType != null){
featureSet.add("ne="+ec.neType);
}
}
Iterator fset = featureSet.iterator();
while (fset.hasNext()) {
string f = (string) fset.next();
features.add(f);
}
*/
}
return features;
}
public override bool CanResolve(MentionContext mention)
{
return(PartsOfSpeech.IsProperNoun(mention.HeadTokenTag) || mention.HeadTokenTag.StartsWith(PartsOfSpeech.CardinalNumber));
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return true;
}
else
{
MentionContext cec = entity.LastExtent;
return (!cec.HeadTokenTag.Equals("NNS") || base.IsExcluded(mention, entity));
}
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return true;
}
foreach (MentionContext xec in entity.Mentions)
{
if (xec.HeadTokenTag.StartsWith("NNP"))
{
// || initialCaps.matcher(xec.headToken.toString()).find()) {
//System.err.println("MaxentProperNounResolver.exclude: kept "+xec.toText()+" with "+xec.headTag);
return false;
}
}
return true;
}
protected internal override bool IsOutOfRange(MentionContext mention, DiscourseEntity entity)
{
var cec = entity.LastExtent;
return(mention.SentenceNumber - cec.SentenceNumber > NumberSentencesBack);
}
public override bool CanResolve(MentionContext mention)
{
return (mention.HeadTokenTag.StartsWith("NNP") || mention.HeadTokenTag.StartsWith("CD"));
}
protected internal virtual List<string> getAcronymFeatures(MentionContext mention, DiscourseEntity entity)
{
MentionContext xec = getProperNounExtent(entity);
string ecStrip = StripNounPhrase(mention);
string xecStrip = StripNounPhrase(xec);
if (ecStrip != null && xecStrip != null)
{
if (isAcronym(ecStrip, xecStrip))
{
List<string> features = new List<string>(1);
features.Add("knownAcronym");
return features;
}
}
return new List<string>();
}
protected internal override bool IsExcluded(MentionContext context, DiscourseEntity discourseEntity)
{
MentionContext currentContext = discourseEntity.LastExtent;
//System.err.println("IsAResolver.excluded?: ec.span="+ec.getSpan()+" cec.span="+cec.getSpan()+" cec="+cec.toText()+" lastToken="+ec.getNextToken());
if (context.SentenceNumber != currentContext.SentenceNumber)
{
//System.err.println("IsAResolver.excluded: (true) not same sentence");
return true;
}
//shallow parse appositives
//System.err.println("IsAResolver.excluded: ec="+ec.toText()+" "+ec.span+" cec="+cec.toText()+" "+cec.span);
if (currentContext.IndexSpan.End == context.IndexSpan.Start - 2)
{
return false;
}
//full parse w/o trailing comma
if (currentContext.IndexSpan.End == context.IndexSpan.End)
{
//System.err.println("IsAResolver.excluded: (false) spans share end");
return false;
}
//full parse w/ trailing comma or period
if (currentContext.IndexSpan.End <= context.IndexSpan.End + 2 && (context.NextToken != null && (context.NextToken.ToString().Equals(",") || context.NextToken.ToString().Equals("."))))
{
//System.err.println("IsAResolver.excluded: (false) spans end + punct");
return false;
}
//System.err.println("IsAResolver.excluded: (true) default");
return true;
}
public override bool CanResolve(MentionContext mention)
{
//System.err.println("MaxentSingularPronounResolver.canResolve: ec= ("+mention.id+") "+ mention.toText());
string tag = mention.HeadTokenTag;
return (tag != null && tag.StartsWith("PRP") && Linker.SingularThirdPersonPronounPattern.IsMatch(mention.HeadTokenText));
}
protected internal override List<string> GetFeatures(MentionContext mention, DiscourseEntity entity)
{
List<string> features = base.GetFeatures(mention, entity);
if (entity != null)
{
MentionContext ant = entity.LastExtent;
List<string> leftContexts = GetContextFeatures(ant);
for (int ci = 0, cn = leftContexts.Count; ci < cn; ci++)
{
features.Add("l" + leftContexts[ci]);
}
List<string> rightContexts = GetContextFeatures(mention);
for (int ci = 0, cn = rightContexts.Count; ci < cn; ci++)
{
features.Add("r" + rightContexts[ci]);
}
features.Add("hts" + ant.HeadTokenTag + "," + mention.HeadTokenTag);
}
/*
if (entity != null) {
//System.err.println("MaxentIsResolver.getFeatures: ["+ec2.toText()+"] -> ["+de.getLastExtent().toText()+"]");
//previous word and tag
if (ant.prevToken != null) {
features.add("pw=" + ant.prevToken);
features.add("pt=" + ant.prevToken.getSyntacticType());
}
else {
features.add("pw=<none>");
features.add("pt=<none>");
}
//next word and tag
if (mention.nextToken != null) {
features.add("nw=" + mention.nextToken);
features.add("nt=" + mention.nextToken.getSyntacticType());
}
else {
features.add("nw=<none>");
features.add("nt=<none>");
}
//modifier word and tag for c1
int i = 0;
List c1toks = ant.tokens;
for (; i < ant.headTokenIndex; i++) {
features.add("mw=" + c1toks.get(i));
features.add("mt=" + ((Parse) c1toks.get(i)).getSyntacticType());
}
//head word and tag for c1
features.add("mh=" + c1toks.get(i));
features.add("mt=" + ((Parse) c1toks.get(i)).getSyntacticType());
//modifier word and tag for c2
i = 0;
List c2toks = mention.tokens;
for (; i < mention.headTokenIndex; i++) {
features.add("mw=" + c2toks.get(i));
features.add("mt=" + ((Parse) c2toks.get(i)).getSyntacticType());
}
//head word and tag for n2
features.add("mh=" + c2toks.get(i));
features.add("mt=" + ((Parse) c2toks.get(i)).getSyntacticType());
//word/tag pairs
for (i = 0; i < ant.headTokenIndex; i++) {
for (int j = 0; j < mention.headTokenIndex; j++) {
features.add("w=" + c1toks.get(i) + "|" + "w=" + c2toks.get(j));
features.add("w=" + c1toks.get(i) + "|" + "t=" + ((Parse) c2toks.get(j)).getSyntacticType());
features.add("t=" + ((Parse) c1toks.get(i)).getSyntacticType() + "|" + "w=" + c2toks.get(j));
features.add("t=" + ((Parse) c1toks.get(i)).getSyntacticType() + "|" + "t=" + ((Parse) c2toks.get(j)).getSyntacticType());
}
}
features.add("ht=" + ant.headTokenTag + "|" + "ht=" + mention.headTokenTag);
features.add("ht1=" + ant.headTokenTag);
features.add("ht2=" + mention.headTokenTag);
*/
//semantic categories
/*
if (ant.neType != null) {
if (re.neType != null) {
features.add("sc="+ant.neType+","+re.neType);
}
else if (!re.headTokenTag.startsWith("NNP") && re.headTokenTag.startsWith("NN")) {
Set synsets = re.synsets;
for (Iterator si=synsets.iterator();si.hasNext();) {
features.add("sc="+ant.neType+","+si.next());
}
}
}
else if (!ant.headTokenTag.startsWith("NNP") && ant.headTokenTag.startsWith("NN")) {
if (re.neType != null) {
Set synsets = ant.synsets;
for (Iterator si=synsets.iterator();si.hasNext();) {
features.add("sc="+re.neType+","+si.next());
}
}
else if (!re.headTokenTag.startsWith("NNP") && re.headTokenTag.startsWith("NN")) {
//System.err.println("MaxentIsaResolover.getFeatures: both common re="+re.parse+" ant="+ant.parse);
Set synsets1 = ant.synsets;
Set synsets2 = re.synsets;
for (Iterator si=synsets1.iterator();si.hasNext();) {
Object synset = si.next();
if (synsets2.contains(synset)) {
features.add("sc="+synset);
}
}
}
}
}
*/
//System.err.println("MaxentIsResolver.getFeatures: "+features.toString());
return features;
}
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return (true);
}
string mentionGender = null;
foreach (MentionContext entityMention in entity.Mentions)
{
string tag = entityMention.HeadTokenTag;
if (tag != null && tag.StartsWith("PRP") && Linker.SingularThirdPersonPronounPattern.IsMatch(mention.HeadTokenText))
{
if (mentionGender == null)
{
//lazy initilization
mentionGender = GetPronounGender(mention.HeadTokenText);
}
string entityGender = GetPronounGender(entityMention.HeadTokenText);
if (!entityGender.Equals("u") && !mentionGender.Equals(entityGender))
{
return (true);
}
}
}
return (false);
}
public override bool CanResolve(MentionContext mention)
{
string tag = mention.HeadTokenTag;
return (tag != null && tag.StartsWith("PRP") && Linker.PluralThirdPersonPronounPattern.IsMatch(mention.HeadTokenText));
}
public override DiscourseEntity Resolve(MentionContext ec, DiscourseModel dm)
{
return (null);
}
protected internal override bool IsOutOfRange(MentionContext ec, DiscourseEntity de)
{
return (false);
}
public override bool CanResolve(MentionContext ec)
{
return (true);
}
protected internal override bool IsOutOfRange(MentionContext mention, DiscourseEntity entity)
{
MentionContext cec = entity.LastExtent;
//System.err.println("MaxentPluralPronounResolver.outOfRange: ["+ec.toText()+" ("+ec.id+")] ["+cec.toText()+" ("+cec.id+")] ec.sentenceNumber=("+ec.sentenceNumber+")-cec.sentenceNumber=("+cec.sentenceNumber+") > "+NUM_SENTS_BACK_PRONOUNS);
return (mention.SentenceNumber - cec.SentenceNumber > NUM_SENTS_BACK_PRONOUNS);
}
protected internal override bool IsOutOfRange(MentionContext mention, DiscourseEntity entity)
{
MentionContext cec = entity.LastExtent;
//System.err.println("MaxentSingularPronounresolve.outOfRange: ["+entity.getLastExtent().toText()+" ("+entity.getId()+")] ["+mention.toText()+" ("+mention.getId()+")] entity.sentenceNumber=("+entity.getLastExtent().getSentenceNumber()+")-mention.sentenceNumber=("+mention.getSentenceNumber()+") > "+numSentencesBack);
return (mention.SentenceNumber - cec.SentenceNumber > NumberSentencesBack);
}
protected internal override bool IsOutOfRange(MentionContext context, DiscourseEntity discourseEntity)
{
MentionContext currentContext = discourseEntity.LastExtent;
return (currentContext.SentenceNumber != context.SentenceNumber);
}
//UPGRADE_NOTE: Access modifiers of method 'excluded' were changed to 'public'. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1204'"
protected internal override bool IsExcluded(MentionContext mention, DiscourseEntity entity)
{
if (base.IsExcluded(mention, entity))
{
return true;
}
MentionContext cec = entity.LastExtent;
if (!CanResolve(cec))
{
return true;
}
if (mention.HeadTokenTag.StartsWith("NNP"))
{
//mention is a propernoun
if (cec.HeadTokenTag.StartsWith("NNP"))
{
return true; // both NNP
}
else
{
if (entity.MentionCount > 1)
{
return true;
}
return !CanResolve(cec);
}
}
else if (mention.HeadTokenTag.StartsWith("PRP"))
{
// mention is a speech pronoun
// cec can be either a speech pronoun or a propernoun
if (cec.HeadTokenTag.StartsWith("NNP"))
{
//exclude antecedents not in the same sentence when they are not pronoun
return (mention.SentenceNumber - cec.SentenceNumber != 0);
}
else if (cec.HeadTokenTag.StartsWith("PRP"))
{
return false;
}
else
{
System.Console.Error.WriteLine("Unexpected candidate exluded: " + cec.ToText());
return true;
}
}
else
{
System.Console.Error.WriteLine("Unexpected mention exluded: " + mention.ToText());
return true;
}
}
protected internal override bool IsOutOfRange(MentionContext mention, DiscourseEntity entity)
{
MentionContext cec = entity.LastExtent;
return (mention.SentenceNumber - cec.SentenceNumber > NumberSentencesBack);
}
