public void TestDisambiguation()
{
var fsm = new FsmMorphologicalAnalyzer();
var corpus = new DisambiguationCorpus("../../../penntreebank.txt");
var algorithm = new LongestRootFirstDisambiguation();
algorithm.Train(corpus);
var correctParse = 0;
var correctRoot = 0;
for (var i = 0; i < corpus.SentenceCount(); i++)
{
var sentenceAnalyses = fsm.RobustMorphologicalAnalysis(corpus.GetSentence(i));
var fsmParses = algorithm.Disambiguate(sentenceAnalyses);
for (var j = 0; j < corpus.GetSentence(i).WordCount(); j++)
{
var word = (DisambiguatedWord)corpus.GetSentence(i).GetWord(j);
if (fsmParses[j].TransitionList().ToLower(new CultureInfo("tr-TR")).Equals(word.GetParse().ToString().ToLower(new CultureInfo("tr-TR"))))
{
correctParse++;
correctRoot++;
}
else
{
if (fsmParses[j].GetWord().Equals(word.GetParse().GetWord()))
{
correctRoot++;
}
}
}
}
Assert.AreEqual(0.9193, (correctRoot + 0.0) / corpus.NumberOfWords(), 0.0001);
Assert.AreEqual(0.8273, (correctParse + 0.0) / corpus.NumberOfWords(), 0.0001);
}
/**
* <summary> Constructor for the class.</summary>
* <param name="rootWordStatistics">The object contains information about the selected correct root words in a corpus for a set
* of possible lemma. For example, the lemma
* `günü': 2 possible root words `gün' and `günü'
* `çağlar' : 2 possible root words `çağ' and `çağlar'</param>
*/
public TurkishSentenceAutoDisambiguator() : base(new FsmMorphologicalAnalyzer())
{
longestRootFirstDisambiguation = new LongestRootFirstDisambiguation();
}
