public static void printTransitionMatrix(Dictionary <StateTransition, double> model)
{
StringBuilder output = new StringBuilder();
WordType[] types = (WordType[])Enum.GetValues(typeof(WordType));
StateTransition trans = new StateTransition(WordType.Undefined, WordType.Undefined);
//Header
output.Append("...,");
for (int n = 0; n < types.Length; n++)
{
output.Append(types[n].ToString() + ",");
}
output.AppendLine();
for (int row = 0; row < types.Length; row++)
{
trans.from = types[row];
output.Append(trans.from.ToString() + ",");
for (int col = 0; col < types.Length; col++)
{
trans.to = types[col];
if (!model.ContainsKey(trans))
{
output.Append("NA" + ",");
}
else
{
output.Append(model[trans].ToString() + ",");
}
}
output.AppendLine();
}
System.IO.File.WriteAllText("transition.csv", output.ToString());
}
static Dictionary <StateTransition, double> createTransitionModel(List <Word> words)
{
Dictionary <StateTransition, int> counter = new Dictionary <StateTransition, int>(); //counts instances of the state transition
Dictionary <WordType, int> fromCounter = new Dictionary <WordType, int>(); //counts times we see the WordType from (needed for percentage)
Dictionary <StateTransition, double> model = new Dictionary <StateTransition, double>(); //Holds answer
for (int i = 0; i < words.Count - 1; i++)
{
if (fromCounter.ContainsKey(words[i].PartOfSpeech))
{
fromCounter[words[i].PartOfSpeech]++;
}
else
{
fromCounter.Add(words[i].PartOfSpeech, 1);
}
StateTransition transition = new StateTransition(words[i].PartOfSpeech, words[i + 1].PartOfSpeech);
if (counter.ContainsKey(transition))
{
counter[transition]++;
}
else
{
counter.Add(transition, 1);
}
}
foreach (KeyValuePair <StateTransition, int> pair in counter)
{
model.Add(pair.Key, (double)pair.Value / (double)fromCounter[(WordType)pair.Key.from]);
}
return(model);
}
