private void InitializeFunctionsArray()
{
for (long i = 0; i < K; i++)
{
_functionsArray[i] = HashFunctionGenerator.Generate(Range);
}
}
/// <summary>Initializes a new instance of the <see cref="LocalitySensitiveHashingCalculator"/> class.</summary>
/// <param name="documents">The documents to compare similarity.</param>
/// <param name="k">The number of MinHash functions.</param>
/// <param name="r">The r value/the row size of b band</param>
public LocalitySensitiveHashingCalculator(IList <Document> documents, int k, int r)
{
Documents = documents;
UniversalSet = documents.SelectMany(d =>
{
var words = d.Words.Select(x => x.Id);
return(words);
}).Distinct().ToArray();
HashFunctions = HashFunctionGenerator.Generate(k, UniversalSet.Length);
_r = r;
}
public double[,] GetMinHashSignatureMatrix(int[,] boolMatrix)
{
HashFunctions = HashFunctionGenerator.Generate(NumberOfHashFunctions, boolMatrix.GetLength(0));
var minHashStopwatch = Stopwatch.StartNew();
double[,] minHashes = new double[HashFunctions.Length, boolMatrix.GetLength(1)];
// Initialise all minhashes as infinity
for (int i = 0; i < minHashes.GetLength(0); i++)
{
for (int j = 0; j < minHashes.GetLength(1); j++)
{
minHashes[i, j] = double.PositiveInfinity;
}
}
// Iterate through boolean matrix
for (int docIndex = 0; docIndex < boolMatrix.GetLength(1); docIndex++)
{
for (int elIndex = 0; elIndex < boolMatrix.GetLength(0); elIndex++)
{
var rowNumber = elIndex + 1;
if (boolMatrix[elIndex, docIndex] == 1)
{
for (int hashFuncIndex = 0; hashFuncIndex < HashFunctions.Length; hashFuncIndex++)
{
var hashValue = HashFunctions[hashFuncIndex](rowNumber);
if (hashValue < minHashes[hashFuncIndex, docIndex])
{
minHashes[hashFuncIndex, docIndex] = hashValue;
}
}
}
}
}
Console.WriteLine($"MinHash Signatures computed in: {minHashStopwatch.ElapsedTicks / TimeSpan.TicksPerMillisecond}ms");
minHashStopwatch.Stop();
return(minHashes);
}
