public void Do_ShouldReturnACollectionOfNGrams_WhenProperArguments()
{
// Arrange
// Act
List <INGram> actual1 = new NGramTokenizer()
.Do(
ObjectMother.NGramTokenizer_Text1,
new TokenizationStrategy(),
new NGramTokenizerRuleSet());
List <INGram> actual2 = new NGramTokenizer()
.Do(
ObjectMother.NGramTokenizer_Text1,
new TokenizationStrategy());
List <INGram> actual3 = new NGramTokenizer()
.Do(
ObjectMother.NGramTokenizer_Text1,
new NGramTokenizerRuleSet());
List <INGram> actual4 = new NGramTokenizer()
.Do(
ObjectMother.NGramTokenizer_Text1);
// Assert
Assert.IsTrue(
ObjectMother.AreEqual(
ObjectMother.NGramTokenizer_Text1_NGrams,
actual1));
Assert.IsTrue(
ObjectMother.AreEqual(
ObjectMother.NGramTokenizer_Text1_NGrams,
actual2));
Assert.IsTrue(
ObjectMother.AreEqual(
ObjectMother.NGramTokenizer_Text1_NGrams,
actual3));
Assert.IsTrue(
ObjectMother.AreEqual(
ObjectMother.NGramTokenizer_Text1_NGrams,
actual4));
}
public void PredictLabel_ShouldReturnANullLabelAndLogTheExpectedMessages_WhenContainsAllIndexAverageValuesEqualToZero()
{
// Arrange
List <string> actualLogMessages = new List <string>();
Action <string> fakeLoggingAction
= (message) => actualLogMessages.Add(message);
TextClassifierComponents components
= new TextClassifierComponents(
new NGramTokenizer(),
new SimilarityIndexCalculatorJaccard(),
TextClassifierComponents.DefaultRoundingFunction,
TextClassifierComponents.DefaultTextTruncatingFunction,
fakeLoggingAction);
TextClassifier textClassifier = new TextClassifier(components, new TextClassifierSettings());
string truncatedText
= TextClassifierComponents.DefaultTextTruncatingFunction.Invoke(
ObjectMother.TextClassifier_Text3,
TextClassifierSettings.DefaultTruncateTextInLogMessagesAfter);
List <string> expectedLogMessages = new List <string>()
{
MessageCollection.TextClassifier_AttemptingToPredictLabel,
MessageCollection.TextClassifier_FollowingTextHasBeenProvided.Invoke(truncatedText),
MessageCollection.TextClassifier_FollowingTokenizationStrategyWillBeUsed.Invoke(new TokenizationStrategy()),
MessageCollection.TextClassifier_FollowingNGramsTokenizerRuleSetWillBeUsed.Invoke(new NGramTokenizerRuleSet()),
MessageCollection.TextClassifier_XLabeledExamplesHaveBeenProvided.Invoke(ObjectMother.TextClassifier_LabeledExamples),
MessageCollection.TextClassifier_ProvidedTextHasBeenTokenizedIntoXNGrams.Invoke(ObjectMother.TextClassifier_Text3_NGrams),
MessageCollection.TextClassifier_ComparingProvidedTextAgainstFollowingLabeledExample.Invoke(ObjectMother.TextClassifier_LabeledExamples[0]),
MessageCollection.TextClassifier_CalculatedSimilarityIndexValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexes[0].Value),
MessageCollection.TextClassifier_RoundedSimilarityIndexValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexes[0].Value),
MessageCollection.TextClassifier_FollowingSimilarityIndexObjectHasBeenAddedToTheList.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexes[0]),
MessageCollection.TextClassifier_ComparingProvidedTextAgainstFollowingLabeledExample.Invoke(ObjectMother.TextClassifier_LabeledExamples[1]),
MessageCollection.TextClassifier_CalculatedSimilarityIndexValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexes[1].Value),
MessageCollection.TextClassifier_RoundedSimilarityIndexValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexes[1].Value),
MessageCollection.TextClassifier_FollowingSimilarityIndexObjectHasBeenAddedToTheList.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexes[1]),
MessageCollection.TextClassifier_TokenizedTextHasBeenComparedAgainstTheProvidedLabeledExamples,
MessageCollection.TextClassifier_XSimilarityIndexObjectsHaveBeenComputed.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexes),
MessageCollection.TextClassifier_FollowingUniqueLabelsHaveBeenFound.Invoke(ObjectMother.TextClassifier_Text3_UniqueLabels),
MessageCollection.TextClassifier_CalculatingIndexAverageForTheFollowingLabel.Invoke(ObjectMother.TextClassifier_Text3_UniqueLabels[0]),
MessageCollection.TextClassifier_CalculatedSimilarityIndexAverageValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexAverages[0].Value),
MessageCollection.TextClassifier_RoundedSimilarityIndexAverageValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexAverages[0].Value),
MessageCollection.TextClassifier_FollowingSimilarityIndexAverageObjectHasBeenAddedToTheList.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexAverages[0]),
MessageCollection.TextClassifier_CalculatingIndexAverageForTheFollowingLabel.Invoke(ObjectMother.TextClassifier_Text3_UniqueLabels[1]),
MessageCollection.TextClassifier_CalculatedSimilarityIndexAverageValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexAverages[1].Value),
MessageCollection.TextClassifier_RoundedSimilarityIndexAverageValueIs.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexAverages[1].Value),
MessageCollection.TextClassifier_FollowingSimilarityIndexAverageObjectHasBeenAddedToTheList.Invoke(ObjectMother.TextClassifier_Text3_SimilarityIndexAverages[1]),
MessageCollection.TextClassifier_XSimilarityIndexAverageObjectsHaveBeenComputed(ObjectMother.TextClassifier_Text3_SimilarityIndexAverages),
MessageCollection.TextClassifier_FollowingVerificationHasFailed.Invoke("ContainsAtLeastOneIndexAverageThatIsNotZero"),
MessageCollection.TextClassifier_PredictedLabelIs.Invoke(ObjectMother.TextClassifier_Text3_Label),
MessageCollection.TextClassifier_PredictionHasFailedTryIncreasingTheAmountOfProvidedLabeledExamples
};
// Act
TextClassifierResult actual
= textClassifier.PredictLabel(
ObjectMother.TextClassifier_Text3,
ObjectMother.TextClassifier_LabeledExamples);
// Assert
Assert.IsTrue(
ObjectMother.AreEqual(
ObjectMother.TextClassifier_Text3_TextClassifierResult,
actual
));
Assert.AreEqual(expectedLogMessages, actualLogMessages);
}