/// <summary>Train a segmenter from raw text.</summary>
/// <remarks>Train a segmenter from raw text. Gold segmentation markers are required.</remarks>
public virtual void Train()
{
bool hasSegmentationMarkers = true;
bool hasTags = true;
IDocumentReaderAndWriter <CoreLabel> docReader = new ArabicDocumentReaderAndWriter(hasSegmentationMarkers, hasTags, hasDomainLabels, domain, noRewrites, tf);
ObjectBank <IList <CoreLabel> > lines = classifier.MakeObjectBankFromFile(flags.trainFile, docReader);
classifier.Train(lines, docReader);
log.Info("Finished training.");
}
/// <summary>
/// Evaluate accuracy when the input is gold segmented text *with* segmentation
/// markers and morphological analyses.
/// </summary>
/// <remarks>
/// Evaluate accuracy when the input is gold segmented text *with* segmentation
/// markers and morphological analyses. In other words, the evaluation file has the
/// same format as the training data.
/// </remarks>
/// <param name="pwOut"/>
private void Evaluate(PrintWriter pwOut)
{
log.Info("Starting evaluation...");
bool hasSegmentationMarkers = true;
bool hasTags = true;
IDocumentReaderAndWriter <CoreLabel> docReader = new ArabicDocumentReaderAndWriter(hasSegmentationMarkers, hasTags, hasDomainLabels, domain, tf);
ObjectBank <IList <CoreLabel> > lines = classifier.MakeObjectBankFromFile(flags.testFile, docReader);
PrintWriter tedEvalGoldTree = null;
PrintWriter tedEvalParseTree = null;
PrintWriter tedEvalGoldSeg = null;
PrintWriter tedEvalParseSeg = null;
if (tedEvalPrefix != null)
{
try
{
tedEvalGoldTree = new PrintWriter(tedEvalPrefix + "_gold.ftree");
tedEvalGoldSeg = new PrintWriter(tedEvalPrefix + "_gold.segmentation");
tedEvalParseTree = new PrintWriter(tedEvalPrefix + "_parse.ftree");
tedEvalParseSeg = new PrintWriter(tedEvalPrefix + "_parse.segmentation");
}
catch (FileNotFoundException e)
{
System.Console.Error.Printf("%s: %s%n", typeof(Edu.Stanford.Nlp.International.Arabic.Process.ArabicSegmenter).FullName, e.Message);
}
}
ICounter <string> labelTotal = new ClassicCounter <string>();
ICounter <string> labelCorrect = new ClassicCounter <string>();
int total = 0;
int correct = 0;
foreach (IList <CoreLabel> line in lines)
{
string[] inputTokens = TedEvalSanitize(IOBUtils.IOBToString(line).ReplaceAll(":", "#pm#")).Split(" ");
string[] goldTokens = TedEvalSanitize(IOBUtils.IOBToString(line, ":")).Split(" ");
line = classifier.Classify(line);
string[] parseTokens = TedEvalSanitize(IOBUtils.IOBToString(line, ":")).Split(" ");
foreach (CoreLabel label in line)
{
// Do not evaluate labeling of whitespace
string observation = label.Get(typeof(CoreAnnotations.CharAnnotation));
if (!observation.Equals(IOBUtils.GetBoundaryCharacter()))
{
total++;
string hypothesis = label.Get(typeof(CoreAnnotations.AnswerAnnotation));
string reference = label.Get(typeof(CoreAnnotations.GoldAnswerAnnotation));
labelTotal.IncrementCount(reference);
if (hypothesis.Equals(reference))
{
correct++;
labelCorrect.IncrementCount(reference);
}
}
}
if (tedEvalParseSeg != null)
{
tedEvalGoldTree.Printf("(root");
tedEvalParseTree.Printf("(root");
int safeLength = inputTokens.Length;
if (inputTokens.Length != goldTokens.Length)
{
log.Info("In generating TEDEval files: Input and gold do not have the same number of tokens");
log.Info(" (ignoring any extras)");
log.Info(" input: " + Arrays.ToString(inputTokens));
log.Info(" gold: " + Arrays.ToString(goldTokens));
safeLength = Math.Min(inputTokens.Length, goldTokens.Length);
}
if (inputTokens.Length != parseTokens.Length)
{
log.Info("In generating TEDEval files: Input and parse do not have the same number of tokens");
log.Info(" (ignoring any extras)");
log.Info(" input: " + Arrays.ToString(inputTokens));
log.Info(" parse: " + Arrays.ToString(parseTokens));
safeLength = Math.Min(inputTokens.Length, parseTokens.Length);
}
for (int i = 0; i < safeLength; i++)
{
foreach (string segment in goldTokens[i].Split(":"))
{
tedEvalGoldTree.Printf(" (seg %s)", segment);
}
tedEvalGoldSeg.Printf("%s\t%s%n", inputTokens[i], goldTokens[i]);
foreach (string segment_1 in parseTokens[i].Split(":"))
{
tedEvalParseTree.Printf(" (seg %s)", segment_1);
}
tedEvalParseSeg.Printf("%s\t%s%n", inputTokens[i], parseTokens[i]);
}
tedEvalGoldTree.Printf(")%n");
tedEvalGoldSeg.Println();
tedEvalParseTree.Printf(")%n");
tedEvalParseSeg.Println();
}
}
double accuracy = ((double)correct) / ((double)total);
accuracy *= 100.0;
pwOut.Println("EVALUATION RESULTS");
pwOut.Printf("#datums:\t%d%n", total);
pwOut.Printf("#correct:\t%d%n", correct);
pwOut.Printf("accuracy:\t%.2f%n", accuracy);
pwOut.Println("==================");
// Output the per label accuracies
pwOut.Println("PER LABEL ACCURACIES");
foreach (string refLabel in labelTotal.KeySet())
{
double nTotal = labelTotal.GetCount(refLabel);
double nCorrect = labelCorrect.GetCount(refLabel);
double acc = (nCorrect / nTotal) * 100.0;
pwOut.Printf(" %s\t%.2f%n", refLabel, acc);
}
if (tedEvalParseSeg != null)
{
tedEvalGoldTree.Close();
tedEvalGoldSeg.Close();
tedEvalParseTree.Close();
tedEvalParseSeg.Close();
}
}
public _ISerializableFunction_131(ArabicDocumentReaderAndWriter _enclosing)
{
this._enclosing = _enclosing;
this.serialVersionUID = 5243251505653686497L;
}
public _ISerializableFunction_131(ArabicDocumentReaderAndWriter _enclosing)
{
this._enclosing = _enclosing;
this.serialVersionUID = serialVersionUID;
}