private double[] GetExactHistories(History h, IList <Pair <int, Extractor> > extractors, IList <Pair <int, Extractor> > extractorsRare)
{
double[] scores = new double[maxentTagger.ySize];
int szCommon = maxentTagger.extractors.Size();
foreach (Pair <int, Extractor> e in extractors)
{
int kf = e.First();
Extractor ex = e.Second();
string val = ex.Extract(h);
int[] fAssociations = maxentTagger.fAssociations[kf][val];
if (fAssociations != null)
{
for (int i = 0; i < maxentTagger.ySize; i++)
{
int fNum = fAssociations[i];
if (fNum > -1)
{
scores[i] += maxentTagger.GetLambdaSolve().lambda[fNum];
}
}
}
}
if (extractorsRare != null)
{
foreach (Pair <int, Extractor> e_1 in extractorsRare)
{
int kf = e_1.First();
Extractor ex = e_1.Second();
string val = ex.Extract(h);
int[] fAssociations = maxentTagger.fAssociations[kf + szCommon][val];
if (fAssociations != null)
{
for (int i = 0; i < maxentTagger.ySize; i++)
{
int fNum = fAssociations[i];
if (fNum > -1)
{
scores[i] += maxentTagger.GetLambdaSolve().lambda[fNum];
}
}
}
}
}
return(scores);
}
public TestSentence(MaxentTagger maxentTagger)
{
// origWords is only set when run with a list of HasWords; when run
// with a list of strings, this will be null
// TODO this always has the value of sent.size(). Remove it? [cdm 2008]
// protected double[][][] probabilities;
// = 0;
System.Diagnostics.Debug.Assert((maxentTagger != null));
System.Diagnostics.Debug.Assert((maxentTagger.GetLambdaSolve() != null));
this.maxentTagger = maxentTagger;
if (maxentTagger.config != null)
{
tagSeparator = maxentTagger.config.GetTagSeparator();
encoding = maxentTagger.config.GetEncoding();
Verbose = maxentTagger.config.GetVerbose();
}
else
{
tagSeparator = TaggerConfig.GetDefaultTagSeparator();
encoding = "utf-8";
Verbose = false;
}
history = new History(pairs, maxentTagger.extractors);
}
internal virtual string ResultsString(MaxentTagger maxentTagger)
{
StringBuilder output = new StringBuilder();
output.Append(string.Format("Model %s has xSize=%d, ySize=%d, and numFeatures=%d.%n", maxentTagger.config.GetModel(), maxentTagger.xSize, maxentTagger.ySize, maxentTagger.GetLambdaSolve().lambda.Length));
output.Append(string.Format("Results on %d sentences and %d words, of which %d were unknown.%n", numSentences, numRight + numWrong, unknownWords));
output.Append(string.Format("Total sentences right: %d (%f%%); wrong: %d (%f%%).%n", numCorrectSentences, numCorrectSentences * 100.0 / numSentences, numSentences - numCorrectSentences, (numSentences - numCorrectSentences) * 100.0 / (numSentences
)));
output.Append(string.Format("Total tags right: %d (%f%%); wrong: %d (%f%%).%n", numRight, numRight * 100.0 / (numRight + numWrong), numWrong, numWrong * 100.0 / (numRight + numWrong)));
if (unknownWords > 0)
{
output.Append(string.Format("Unknown words right: %d (%f%%); wrong: %d (%f%%).%n", (unknownWords - numWrongUnknown), 100.0 - (numWrongUnknown * 100.0 / unknownWords), numWrongUnknown, numWrongUnknown * 100.0 / unknownWords));
}
return(output.ToString());
}
