// Use this for initialization
void Start()
{
// Initialize N-Gram
predictor = useHierarchicalNGram
? new HierarchNGram<char>(nValue, threshold) as INGram<char>
: new NGram<char>(nValue) as INGram<char>;
// Initialize list of chars
listOfChars = new List<char>();
// Initialize right and wrong predictions
right = 0;
wrong = 0;
// Initialize prediction to none
prediction = default(char);
}
public static INGram GetNGram(this INGram ingram)
{
switch (ingram.GetType().ToString().ToLower())
{
case "unigram":
return(new Unigram());
case "filteredunigram":
return(new FilteredUnigram());
case "bigram":
return(new Bigram(new Unigram()));
case "filteredbigram":
return(new Bigram(new FilteredUnigram()));
default:
return(null);
}
}
public Classifier(FileInfo trainedNetwork, Characteristic characteristic,
List <VocabularyWord> vocabulary, INGram inGram)
{
if (characteristic == null ||
characteristic.Name.Equals("") ||
characteristic.PossibleValues == null ||
characteristic.PossibleValues.Count == 0 ||
vocabulary == null ||
vocabulary.Count == 0 ||
inGram == null)
{
throw new ArgumentException();
}
this._characteristic = characteristic;
this._vocabulary = vocabulary;
_inputLayerSize = vocabulary.Count;
_outputLayerSize = characteristic.PossibleValues.Count;
this._inGram = inGram;
if (trainedNetwork == null)
{
_network = CreateNeuralNetwork();
}
else
{
try
{
_network = (BasicNetwork)EncogDirectoryPersistence.LoadObject(trainedNetwork);
}
catch (PersistError)
{
throw new ArgumentException();
}
}
}
public Classifier(Characteristic characteristic,
List <VocabularyWord> vocabulary, INGram inGram) : this(null, characteristic, vocabulary,
inGram)
{
}
public VocabularyBuilder(INGram inGram)
{
_inGram = inGram ?? throw new ArgumentNullException();
}
public Bigram(INGram nGram)
{
_nGram = nGram ?? throw new ArgumentNullException();
}
