/// <summary>
/// Initializes a new instance of the <see cref="DocumentCategorizerME"/> with a document categorizer model.
/// The default feature generation will be used.
/// </summary>
/// <param name="model">The model.</param>
/// <exception cref="System.ArgumentNullException">model</exception>
public DocumentCategorizerME(DocumentCategorizerModel model) {
if (model == null)
throw new ArgumentNullException("model");
cg = new DocumentCategorizerContextGenerator(model.Factory.FeatureGenerators);
this.model = model;
}
/// <summary>
/// Initializes a new instance of the <see cref="DocumentCategorizerME"/> with a document categorizer model.
/// The default feature generation will be used.
/// </summary>
/// <param name="model">The model.</param>
/// <exception cref="System.ArgumentNullException">model</exception>
public DocumentCategorizerME(DocumentCategorizerModel model)
{
if (model == null)
{
throw new ArgumentNullException("model");
}
cg = new DocumentCategorizerContextGenerator(model.Factory.FeatureGenerators);
this.model = model;
}
public void TestDefault() {
var model = Train();
Assert.NotNull(model);
using (var data = new MemoryStream()) {
model.Serialize(new UnclosableStream(data));
data.Seek(0, SeekOrigin.Begin);
var deserialized = new DocumentCategorizerModel(data);
Assert.NotNull(deserialized);
Assert.NotNull(deserialized.Factory);
Assert.AreEqual(1, deserialized.Factory.FeatureGenerators.Length);
Assert.AreEqual(typeof (BagOfWordsFeatureGenerator), deserialized.Factory.FeatureGenerators[0].GetType());
Assert.AreEqual(typeof (WhitespaceTokenizer), deserialized.Factory.Tokenizer.GetType());
}
}
public void TestCustom() {
var featureGenerators = new IFeatureGenerator[] {
new BagOfWordsFeatureGenerator(),
new NGramFeatureGenerator()
};
var factory = new DocumentCategorizerFactory(SimpleTokenizer.Instance, featureGenerators);
var model = Train(factory);
Assert.NotNull(model);
using (var data = new MemoryStream()) {
model.Serialize(new UnclosableStream(data));
data.Seek(0, SeekOrigin.Begin);
var deserialized = new DocumentCategorizerModel(data);
Assert.NotNull(deserialized);
Assert.NotNull(deserialized.Factory);
Assert.AreEqual(2, deserialized.Factory.FeatureGenerators.Length);
Assert.AreEqual(typeof (BagOfWordsFeatureGenerator), deserialized.Factory.FeatureGenerators[0].GetType());
Assert.AreEqual(typeof (NGramFeatureGenerator), deserialized.Factory.FeatureGenerators[1].GetType());
Assert.AreEqual(typeof (SimpleTokenizer), deserialized.Factory.Tokenizer.GetType());
}
}