private static void test(WordAligner wordAligner, List<SentencePair> testSentencePairs, Map<Integer, Alignment> testAlignments, boolean verbose) {
int proposedSureCount = 0;
int proposedPossibleCount = 0;
int sureCount = 0;
int proposedCount = 0;
for (SentencePair sentencePair : testSentencePairs) {
Alignment proposedAlignment = wordAligner.alignSentencePair(sentencePair);
Alignment referenceAlignment = testAlignments.get(sentencePair.getSentenceID());
if (referenceAlignment == null)
throw new RuntimeException("No reference alignment found for sentenceID "+sentencePair.getSentenceID());
if (verbose) System.out.println("Alignment:\n"+Alignment.render(referenceAlignment,proposedAlignment,sentencePair));
for (int frenchPosition = 0; frenchPosition < sentencePair.getFrenchWords().size(); frenchPosition++) {
for (int englishPosition = 0; englishPosition < sentencePair.getEnglishWords().size(); englishPosition++) {
boolean proposed = proposedAlignment.containsSureAlignment(englishPosition, frenchPosition);
boolean sure = referenceAlignment.containsSureAlignment(englishPosition, frenchPosition);
boolean possible = referenceAlignment.containsPossibleAlignment(englishPosition, frenchPosition);
if (proposed && sure) proposedSureCount += 1;
if (proposed && possible) proposedPossibleCount += 1;
if (proposed) proposedCount += 1;
if (sure) sureCount += 1;
}
}
}
System.out.println("Precision: "+proposedPossibleCount/(double)proposedCount);
System.out.println("Recall: "+proposedSureCount/(double)sureCount);
System.out.println("AER: "+(1.0-(proposedSureCount+proposedPossibleCount)/(double)(sureCount+proposedCount)));
}
public static String render(Alignment reference, Alignment proposed, SentencePair sentencePair) {
StringBuilder sb = new StringBuilder();
for (int frenchPosition = 0; frenchPosition < sentencePair.getFrenchWords().size(); frenchPosition++) {
for (int englishPosition = 0; englishPosition < sentencePair.getEnglishWords().size(); englishPosition++) {
boolean sure = reference.containsSureAlignment(englishPosition, frenchPosition);
boolean possible = reference.containsPossibleAlignment(englishPosition, frenchPosition);
char proposedChar = ' ';
if (proposed.containsSureAlignment(englishPosition, frenchPosition))
proposedChar = '#';
if (sure) {
sb.append('[');
sb.append(proposedChar);
sb.append(']');
} else {
if (possible) {
sb.append('(');
sb.append(proposedChar);
sb.append(')');
} else {
sb.append(' ');
sb.append(proposedChar);
sb.append(' ');
}
}
}
sb.append("| ");
sb.append(sentencePair.getFrenchWords().get(frenchPosition));
sb.append('\n');
}
for (int englishPosition = 0; englishPosition < sentencePair.getEnglishWords().size(); englishPosition++) {
sb.append("---");
}
sb.append("'\n");
boolean printed = true;
int index = 0;
while (printed) {
printed = false;
StringBuilder lineSB = new StringBuilder();
for (int englishPosition = 0; englishPosition < sentencePair.getEnglishWords().size(); englishPosition++) {
String englishWord = sentencePair.getEnglishWords().get(englishPosition);
if (englishWord.length() > index) {
printed = true;
lineSB.append(' ');
lineSB.append(englishWord.charAt(index));
lineSB.append(' ');
} else {
lineSB.append(" ");
}
}
index += 1;
if (printed) {
sb.append(lineSB);
sb.append('\n');
}
}
return sb.toString();
}