public void VerifyRemovesInvalidTerms()
{
var termsCollection = new TermsCollection(new[] {"accepted"});
var document = new Document(new[] {"accepted", "nonaccepted", "accepted"});
var documentPurer = new DocumentPurer(termsCollection);
Document puredDocument = documentPurer.PureDocument(document);
CollectionAssert.AreEqual(new[] {"accepted", "accepted"}, puredDocument.Words);
}
public void DoesNotAddInexistingTermToTermsCollection()
{
var document1 = new Document(new[] { "first", "word" });
var document2 = new Document(new[] { "word", "second" });
var termsCollection = new TermsCollection();
termsCollection.AddDocument(document1);
termsCollection.AddDocument(document2);
Assert.That(termsCollection.ContainsTerm("nonexisting"), Is.False);
}
public void ContainsAllWordsInPublicCollection()
{
var document1 = new Document(new[] { "first", "word" });
var document2 = new Document(new[] { "word", "second" });
var termsCollection = new TermsCollection();
termsCollection.AddDocument(document1);
termsCollection.AddDocument(document2);
CollectionAssert.AreEqual(new[] { "first", "word", "second"}, termsCollection.Terms);
}
public void AddsAllTermsFromDocumentsToTermsCollection()
{
var document1 = new Document(new[] {"first", "word"});
var document2 = new Document(new[] {"word", "second"});
var termsCollection = new TermsCollection();
termsCollection.AddDocument(document1);
termsCollection.AddDocument(document2);
Assert.That(termsCollection.ContainsTerm("first"), Is.True);
Assert.That(termsCollection.ContainsTerm("word"), Is.True);
Assert.That(termsCollection.ContainsTerm("second"), Is.True);
}
public void DistanceCalculation()
{
var termWeightRepresentation = new Mock<ITermWeightRepresentation>();
termWeightRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "bee"))).Returns(0.91629);
termWeightRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "wasp"))).Returns(0.91629);
termWeightRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "fly"))).Returns(0.22314);
var termsCollection = new TermsCollection(new[] {"bee", "wasp", "fly", "fruit", "like"});
var distanceCalculator = new DistanceCalculator();
double actualDistance = distanceCalculator.CalculateDistance(termWeightRepresentation.Object,
termsCollection);
Assert.AreEqual(1.314903211, actualDistance, 0.0001);
}
public double CalculateProbability(ITermWeightRepresentation queryRepresentation,
ITermWeightRepresentation documentRepresentation,
TermsCollection termsCollection)
{
double queryDistance = _distanceCalculator.CalculateDistance(queryRepresentation, termsCollection);
double documentDistance = _distanceCalculator.CalculateDistance(documentRepresentation, termsCollection);
double sum = 0;
foreach (string term in termsCollection.Terms)
{
sum += queryRepresentation.TermWeight(term)*documentRepresentation.TermWeight(term);
}
return sum/(queryDistance*documentDistance);
}
public void ProbabilityMatrixCalculation()
{
var queryRepresentation = new Mock<ITermWeightRepresentation>();
var documentRepresentation = new Mock<ITermWeightRepresentation>();
queryRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "fruit"))).Returns(0.51083);
queryRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "fly"))).Returns(0.22314);
documentRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "bee"))).Returns(0.91629);
documentRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "wasp"))).Returns(0.91629);
documentRepresentation.Setup(x => x.TermWeight(It.Is<string>(y => y == "fly"))).Returns(0.22314);
var termsCollection = new TermsCollection(new[] {"bee", "wasp", "fly", "fruit", "like"});
var distanceCalculator = new DistanceCalculator();
var probabilityMatrix = new ProbabilityMatrixCalculator();
double probability = probabilityMatrix.CalculateProbability(queryRepresentation.Object,
documentRepresentation.Object, termsCollection);
Assert.AreEqual(0.067932752, probability, 0.0001);
}
