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);
}
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 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)
{
Object[] mtokens = mention.Tokens;
string firstTok = mention.FirstTokenText.ToLower();
bool rv = mtokens.Length > 1 && !PartsOfSpeech.IsProperNoun(mention.HeadTokenTag) && IsDefiniteArticle(firstTok, mention.FirstTokenTag);
return(rv);
}
//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);
}
}
private MentionContext getProperNounExtent(DiscourseEntity de)
{
foreach (MentionContext xec in de.Mentions)
{
//use first extent which is propername
string xecHeadTag = xec.HeadTokenTag;
if (PartsOfSpeech.IsProperNoun(xecHeadTag) || InitialCaps.IsMatch(xec.HeadTokenText))
{
return(xec);
}
}
return(null);
}
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);
}
private void Initialize(Mention.IHeadFinder headFinder)
{
Mention.IParse head = headFinder.GetLastHead(Parse);
List <Mention.IParse> tokenList = head.Tokens;
this.HeadTokenIndex = headFinder.GetHeadIndex(head);
Mention.IParse headToken = headFinder.GetHeadToken(head);
_tokens = tokenList.ToArray();
this.HeadTokenTag = headToken.SyntacticType;
this.HeadTokenText = headToken.ToString();
if (PartsOfSpeech.IsNoun(this.HeadTokenTag) && !PartsOfSpeech.IsProperNoun(this.HeadTokenTag))
{
this.Synsets = GetSynsetSet(this);
}
else
{
this.Synsets = new Util.HashSet <string>();
}
}
public override bool CanResolve(MentionContext mention)
{
return(PartsOfSpeech.IsProperNoun(mention.HeadTokenTag) || mention.HeadTokenTag.StartsWith(PartsOfSpeech.CardinalNumber));
}
private bool IsCommonNoun(Context nounPhrase)
{
return(!PartsOfSpeech.IsProperNoun(nounPhrase.HeadTokenTag) && PartsOfSpeech.IsNoun(nounPhrase.HeadTokenTag));
}
private bool IsName(Context nounPhrase)
{
return(PartsOfSpeech.IsProperNoun(nounPhrase.HeadTokenTag));
}
/// <summary>
/// Returns a string for the specified mention with punctuation, honorifics, designators, and determiners removed.
/// </summary>
/// <param name="mention">
/// The mention to be stripped.
/// </param>
/// <returns>
/// a normalized string representation of the specified mention.
/// </returns>
protected internal virtual string StripNounPhrase(Mention.MentionContext mention)
{
int start = mention.NonDescriptorStart; //start after descriptors
Mention.IParse[] mentionTokens = mention.TokenParses;
int end = mention.HeadTokenIndex + 1;
if (start == end)
{
return(null);
}
//strip determiners
if (mentionTokens[start].SyntacticType == PartsOfSpeech.Determiner)
{
start++;
}
if (start == end)
{
return(null);
}
//get to first NNP
string type;
for (int index = start; index < end; index++)
{
type = mentionTokens[start].SyntacticType;
if (PartsOfSpeech.IsProperNoun(type))
{
break;
}
start++;
}
if (start == end)
{
return(null);
}
if (start + 1 != end)
{
// don't do this on head words, to keep "U.S."
//strip off honorifics in begining
if (Linker.HonorificsPattern.IsMatch(mentionTokens[start].ToString()))
{
start++;
}
if (start == end)
{
return(null);
}
//strip off and honorifics on the end
if (Linker.DesignatorsPattern.IsMatch(mentionTokens[mentionTokens.Length - 1].ToString()))
{
end--;
}
}
if (start == end)
{
return(null);
}
var strip = new StringBuilder();
for (int i = start; i < end; i++)
{
strip.Append(mentionTokens[i].ToString()).Append(" ");
}
return(strip.ToString().Trim());
}