public static double CalcCosinusBetweenWordsCounts(
WordsAndCounts wordsCounts1, WordsAndCounts wordsCounts2)
{
// Create a list (union) of all words from the two sets of words
Dictionary <string, bool> allWords = new Dictionary <string, bool>();
foreach (string word in wordsCounts1.Words)
{
allWords.Add(word, true);
}
foreach (string word in wordsCounts2.Words)
{
if (!allWords.ContainsKey(word))
{
allWords.Add(word, true);
}
}
// Create the first occurences vector
double[] vector1 = new double[allWords.Count];
int index1 = 0;
foreach (string word in allWords.Keys)
{
vector1[index1] = wordsCounts1[word];
index1++;
}
// Create the second occurences vector
double[] vector2 = new double[allWords.Count];
int index2 = 0;
foreach (string word in allWords.Keys)
{
vector2[index2] = wordsCounts2[word];
index2++;
}
double distance = VectorUtils.CalcCosinusBetweenVectors(vector1, vector2);
return(distance);
}
/// <summary>
/// Calculates the distance (similarity) between given Bulgarian and Russian words.
/// </summary>
public static double CrossSim(string bgWord, string ruWord)
{
double[] bgVector = CalculateBgDictionaryContextVector(bgWord);
double[] ruVector = CalculateRuDictionaryContextVector(ruWord);
double distance;
if (vectorDiffAlgorithm == VectorDiffAlgorithm.COSINE)
{
distance = VectorUtils.CalcCosinusBetweenVectors(bgVector, ruVector);
}
else if (vectorDiffAlgorithm == VectorDiffAlgorithm.DICE_COEFFICIENT)
{
distance = VectorUtils.CalcDiceCoeffBetweenVectors(bgVector, ruVector);
}
else
{
throw new Exception("Invalid vector diff algorithm!");
}
return(distance);
}