public void Split(string name, double minAccuracy, Sentence[] train, Sentence[] test)
{
Assert.Greater(train.Length, 0);
Assert.Greater(test.Length, 0);
Assert.Greater(train.Length, test.Length);
LuMiiTagger tagger = new LuMiiTagger();
Stopwatch trainTimer = new Stopwatch();
trainTimer.Start();
tagger.Train(train);
trainTimer.Stop();
Stopwatch tagTimer = new Stopwatch();
tagTimer.Start();
tagger.Tag(test);
tagTimer.Stop();
Token[] tokens = test.SelectMany(t => t).ToArray();
double accuracyTag = (double)tokens.Count(t => t.IsTagCorrect) / tokens.Count();
double accuracyMsd = (double)tokens.Count(t => t.IsMsdCorrect) / tokens.Count();
double accuracyMsdLemma = (double)tokens.Count(t => t.IsMsdCorrect && t.IsLemmaCorrect) / tokens.Count();
double accuracyLemma = (double)tokens.Count(t => t.IsLemmaCorrect) / tokens.Count();
Token[] lemmaIncorrect = tokens.Where(t => t.IsMsdCorrect && !t.IsLemmaCorrect).ToArray();
Debug.WriteLine("Split validation for " + name);
Debug.WriteLine("Train: {0} sentences, {1} tokens", train.Length, train.SelectMany(t => t).Count());
Debug.WriteLine("Test: {0} sentences, {1} tokens", test.Length, test.SelectMany(t => t).Count());
Debug.WriteLine("Accuracy tag: {0:0.00}%", accuracyTag * 100);
Debug.WriteLine("Accuracy msd: {0:0.00}%", accuracyMsd * 100);
Debug.WriteLine("Accuracy msd + lemma: {0:0.00}%", accuracyMsdLemma * 100);
Debug.WriteLine("Accuracy lemma: {0:0.00}%", accuracyLemma * 100);
Debug.WriteLine("Train duration: {0} or {1:0} ms", trainTimer.Elapsed, trainTimer.ElapsedMilliseconds);
Debug.WriteLine("Tag duration: {0} or {1:0} ms", tagTimer.Elapsed, tagTimer.ElapsedMilliseconds);
Debug.WriteLine("Tag speed: {0:0.00} tokens/s", tokens.Length / tagTimer.Elapsed.TotalSeconds);
Assert.Greater(accuracyMsdLemma, minAccuracy);
Assert.Less(accuracyMsdLemma, 0.97);
}
public void TagSpeed()
{
string trainResource = Analyzed2Train;
string testResource = Analyzed2Test;
int maxTokenCount = 1000000;
double minAccuracy = 0.93;
Sentence[] train = LoadAnalyzedCorpus(trainResource);
Sentence[] test = LoadAnalyzedCorpus(testResource);
List<Sentence> all = new List<Sentence>();
int allTokenCount = 0;
while (allTokenCount < maxTokenCount)
{
Sentence s = new Sentence();
foreach (Sentence sentence in test)
{
foreach (Token token in sentence)
{
s.Add(new Token(token));
allTokenCount++;
if (allTokenCount >= maxTokenCount) break;
}
if (allTokenCount >= maxTokenCount) break;
}
all.Add(s);
if (allTokenCount >= maxTokenCount) break;
}
Assert.AreEqual(maxTokenCount, all.SelectMany(t => t).Count());
LuMiiTagger tagger = new LuMiiTagger();
tagger.Train(train);
Stopwatch timer = new Stopwatch();
timer.Start();
tagger.Tag(all);
timer.Stop();
Token[] tokens = all.SelectMany(t => t).ToArray();
double accuracy = (double)tokens.Count(t => t.IsTagCorrect) / tokens.Count();
Assert.AreEqual(maxTokenCount, tokens.Length);
Debug.WriteLine("Accuracy: {0:0.00}%", accuracy * 100);
Debug.WriteLine("Tokens: {0}%", tokens.Length);
Debug.WriteLine("Tag duration: {0} or {1:0} ms", timer.Elapsed, timer.ElapsedMilliseconds);
Debug.WriteLine("Tag speed: {0:0.00} tokens/s", tokens.Length / timer.Elapsed.TotalSeconds);
Assert.Greater(accuracy, minAccuracy);
Assert.Less(accuracy, 0.97);
}
public void LoadSave()
{
string trainResource = Analyzed2Train;
string testResource = Analyzed2Test;
double minAccuracy = 0.93;
Sentence[] train = LoadAnalyzedCorpus(trainResource);
Sentence[] test = LoadAnalyzedCorpus(testResource);
Sentence[] test2 = LoadAnalyzedCorpus(testResource);
Assert.Greater(train.Length, 0);
Assert.Greater(test.Length, 0);
Assert.Greater(train.Length, test.Length);
string filename = Path.GetTempFileName();
LuMiiTagger tagger = new LuMiiTagger();
tagger.Train(train);
tagger.Save(filename);
tagger.Tag(test);
LuMiiTagger tagger2 = new LuMiiTagger();
tagger2.Load(filename);
tagger2.Tag(test2);
Token[] tokens = test.SelectMany(t => t).ToArray();
Token[] tokens2 = test2.SelectMany(t => t).ToArray();
double accuracy = (double)tokens.Count(t => t.IsTagCorrect) / tokens.Count();
double accuracy2 = (double)tokens2.Count(t => t.IsTagCorrect) / tokens2.Count();
Assert.AreEqual(accuracy, accuracy2, 0.0001);
Assert.Greater(accuracy, minAccuracy);
Assert.Less(accuracy, 0.97);
File.Delete(filename);
}