public List<KeyValuePair<string, double>> RankedResults()
{
List<Document> documents = new List<Document>();
HashSet<string> dataSet = new HashSet<string>();
foreach (var result in results)
{
Document d = new Document(result);
documents.Add(d);
foreach (var term in d.tokens()){
dataSet.Add(term);
}
}
//Build Document Vectors
Dictionary<string, Vector> documentVectors = new Dictionary<string, Vector>();
foreach (var document in documents)
{
documentVectors.Add(document.ToString(), new Vector(dataSet, document));
}
//Build Query Vector
Query query = new Query(queryString);
Vector queryVector = new Vector(dataSet, query);
Dictionary<string, double> relevance = new Dictionary<string, double>();
foreach (var documentVector in documentVectors)
{
relevance.Add(documentVector.Key, Vector.GetSimilarityScore(queryVector, documentVector.Value));
}
//Sort result by most relevant
List<KeyValuePair<string, double>> myList = relevance.ToList();
return myList;
}
public void TestGetVectorNorm1()
{
double[] vectorList1 = { 0, 0, 0, 0, 0, 0, 0 };
var vector1 = new Vector(vectorList1);
double expected_result = 0;
double actual_result = Vector.GetVectorNorm(vector1);
Assert.AreEqual(expected_result, actual_result, "Test GetVectorNorm1 failed");
}
public static double GetVectorNorm(Vector vector)
{
double result = 0.0;
foreach (var weight in vector.vectorRep)
{
result += weight * weight;
}
return result;
}
public static double GetDotProduct(Vector vectorA, Vector vectorB)
{
double result = 0.0;
for (int i = 0; i < vectorA.vectorRep.Count; i++)
{
result += vectorA.vectorRep[i] * vectorB.vectorRep[i];
}
return result;
}
public void TestGetSimilarityScore2()
{
double[] vectorList1 = { 0, 0, 0, 0, 0, 0, 0 };
double[] vectorList2 = { 1.12, 0.91, 1.414, 0, 0.89, 0, 1.0 };
var vector1 = new Vector(vectorList1);
var vector2 = new Vector(vectorList2);
double expected_result = 0;
double actual_result = Vector.GetSimilarityScore(vector1, vector2);
Assert.AreEqual(expected_result, actual_result, "Test GetSimilarityScore2 failed");
}
public void TestGetSimilarityScore1()
{
double[] vectorList1 = { 1.0, 0.89, 0, 1.0, 2.14, 3.414, 1.414 };
double[] vectorList2 = { 1.12, 0.91, 1.414, 0, 0.89, 0, 1.0 };
var vector1 = new Vector(vectorList1);
var vector2 = new Vector(vectorList2);
double expected_result = Math.Cos(5.2485/(5.873996 * 21.026492));
double actual_result = Vector.GetSimilarityScore(vector1, vector2);
Assert.AreEqual(expected_result, actual_result, "Test GetSimilarityScore1 failed");
}
public void TestGetDotProduct3()
{
double[] vectorList1 = { 0, 0, 0, 0, 0, 0, 0 };
double[] vectorList2 = { 0, 0, 0, 0, 0, 0, 0 };
var vector1 = new Vector(vectorList1);
var vector2 = new Vector(vectorList2);
double expected_result = 0;
double actual_result = Vector.GetDotProduct(vector1, vector2);
Assert.AreEqual(expected_result, actual_result, "Test GetDotProduct3 failed");
}
public void TestGetDotProduct1()
{
double[] vectorList1 = {1.0, 0.89, 0, 1.0, 2.14, 3.414, 1.414};
double[] vectorList2 = { 1.12, 0.91, 1.414, 0, 0.89, 0, 1.0 };
var vector1 = new Vector(vectorList1);
var vector2 = new Vector(vectorList2);
double expected_result = 5.2485;
double actual_result = Vector.GetDotProduct(vector1, vector2);
Assert.AreEqual(expected_result, actual_result, "Test GetDotProduct1 failed");
}
//compares two vectors and returns their similarity using cosine similarity model.
public static double GetSimilarityScore(Vector vectorA, Vector vectorB)
{
double dotProduct = GetDotProduct(vectorA, vectorB);
if (dotProduct == 0)
{
return 0;
}
else
{
return Math.Cos(dotProduct / ((GetVectorNorm(vectorA)) * GetVectorNorm(vectorB)));
}
}
public void TestGetVectorNorm2()
{
double[] vectorList1 = { 1.0, 0.89, 0, 1.0, 2.14, 3.414, 1.414 };
var vector1 = new Vector(vectorList1);
double expected_result = 21.026492;
double actual_result = Vector.GetVectorNorm(vector1);
Assert.AreEqual(expected_result, actual_result, "Test GetVectorNorm2 failed");
}