internal override string Extract(History h, PairsHolder pH)
{
// I ran a bunch of timing tests that seem to indicate it is
// cheaper to simply add string + char + string than use a
// StringBuilder or go through the StringBuildMemoizer -horatio
return(pH.GetWord(h, leftPosition) + '!' + pH.GetWord(h, rightPosition));
}
public virtual void PrintSent()
{
Print(System.Console.Out);
for (int i = this.start; i < this.end; i++)
{
System.Console.Out.Write(pairs.GetTag(i) + ' ' + pairs.GetWord(i) + '\t');
}
System.Console.Out.WriteLine();
}
internal override string Extract(History h, PairsHolder pH)
{
string cword = pH.GetWord(h, 0);
int allCount = dict.Sum(cword);
int vBNCount = dict.GetCount(cword, vbnTag);
int vBDCount = dict.GetCount(cword, vbdTag);
// Conditions for deciding inapplicable
if ((allCount == 0) && (!(cword.EndsWith(edSuff) || cword.EndsWith(enSuff))))
{
return(zeroSt);
}
if ((allCount > 0) && (vBNCount + vBDCount <= allCount / 100))
{
return(zeroSt);
}
string lastverb = naWord;
//String lastvtag = zeroSt; // mg: written but never read
for (int index = -1; index >= -bound; index--)
{
string word2 = pH.GetWord(h, index);
if ("NA".Equals(word2))
{
break;
}
if (stopper.Matcher(word2).Matches())
{
break;
}
if (vbnWord.Matcher(word2).Matches())
{
lastverb = word2;
break;
}
index--;
}
if (!lastverb.Equals(naWord))
{
log.Info("VBN: For " + cword + ", found preceding VBN cue " + lastverb);
return(oneSt);
}
return(zeroSt);
}
internal override string Extract(History h, PairsHolder pH)
{
string cw = pH.GetWord(h, 0);
string lk = cw.ToLower(Locale.English);
if (lk.Equals(cw))
{
return(zeroSt);
}
return(cw);
}
internal override string Extract(History h, PairsHolder pH)
{
StringBuilder sb = new StringBuilder();
for (int j = left; j <= right; j++)
{
string s = pH.GetWord(h, j);
sb.Append(WordShapeClassifier.WordShape(s, wordShaper));
if (j < right)
{
sb.Append('|');
}
}
return(sb.ToString());
}
internal override string Extract(History h, PairsHolder pH)
{
StringBuilder sb = new StringBuilder();
for (int j = left; j <= right; j++)
{
string word = pH.GetWord(h, j);
string distSim = lexicon.GetMapping(word);
sb.Append(distSim);
if (j < right)
{
sb.Append('|');
}
}
return(sb.ToString());
}
// This precomputes scores of local features (localScores).
protected internal virtual double[] GetHistories(string[] tags, History h)
{
bool rare = maxentTagger.IsRare(ExtractorFrames.cWord.Extract(h));
Extractors ex = maxentTagger.extractors;
Extractors exR = maxentTagger.extractorsRare;
string w = pairs.GetWord(h.current);
double[] lS;
double[] lcS;
lS = localScores[w];
if (lS == null)
{
lS = GetHistories(tags, h, ex.local, rare ? exR.local : null);
localScores[w] = lS;
}
else
{
if (lS.Length != tags.Length)
{
// This case can occur when a word was given a specific forced
// tag, and then later it shows up without the forced tag.
// TODO: if a word is given a forced tag, we should always get
// its features rather than use the cache, just in case the tag
// given is not the same tag as before
lS = GetHistories(tags, h, ex.local, rare ? exR.local : null);
if (tags.Length > 1)
{
localScores[w] = lS;
}
}
}
if ((lcS = localContextScores[h.current]) == null)
{
lcS = GetHistories(tags, h, ex.localContext, rare ? exR.localContext : null);
localContextScores[h.current] = lcS;
ArrayMath.PairwiseAddInPlace(lcS, lS);
}
double[] totalS = GetHistories(tags, h, ex.dynamic, rare ? exR.dynamic : null);
ArrayMath.PairwiseAddInPlace(totalS, lcS);
return(totalS);
}
internal virtual string ExtractLV(History h, PairsHolder pH, int bound)
{
// should extract last verbal word and also the current word
int start = h.start;
string lastverb = "NA";
int current = h.current;
int index = current - 1;
while ((index >= start) && (index >= current - bound))
{
string tag = pH.GetTag(index);
if (tag.StartsWith("VB"))
{
lastverb = pH.GetWord(index);
break;
}
if (tag.StartsWith(","))
{
break;
}
index--;
}
return(lastverb);
}
internal virtual string Extract(History h, PairsHolder pH)
{
return(isTag ? pH.GetTag(h, position) : pH.GetWord(h, position));
}
internal override string Extract(History h, PairsHolder pH)
{
return(pH.GetTag(h, position1) + '!' + pH.GetWord(h, word) + '!' + pH.GetTag(h, position2));
}
internal override string Extract(History h, PairsHolder pH)
{
return(pH.GetWord(h, leftWord) + '!' + pH.GetTag(h, tag) + '!' + pH.GetWord(h, rightWord));
}
internal override string Extract(History h, PairsHolder pH)
{
return(pH.GetWord(h, position).ToLower(Locale.English));
}