/// <summary>
/// Get hash similarity of one song
/// </summary>
/// <param name = "service">Fingerprint service</param>
/// <param name = "hashTables">Number of hash tables in the LSH transformation</param>
/// <param name = "hashKeys">Number of hash keys per table in the LSH transformation</param>
/// <param name = "path">Path to analyzed file</param>
/// <param name = "results">Results object to be filled with the appropriate data</param>
private void GetHashSimilarity(IFingerprintService service, int hashTables, int hashKeys, IWorkUnit unitOfWork, IWorkUnit sameUnitOfWork, DumpResults results)
{
double sum = 0;
int hashesCount = 0;
int startindex = 0;
List<bool[]> listDb = unitOfWork.GetFingerprintsUsingService(service).Result;
List<bool[]> listQuery = sameUnitOfWork.GetFingerprintsUsingService(service).Result;
IPermutations perms = new DbPermutations(ConfigurationManager.ConnectionStrings["FingerprintConnectionString"].ConnectionString);
MinHash minHash = new MinHash(perms);
List<int[]> minHashDb = listDb.Select(minHash.ComputeMinHashSignature).ToList();
List<int[]> minHashQuery = listQuery.Select(minHash.ComputeMinHashSignature).ToList();
/*Calculate Min Hash signature similarity by comparing 2 consecutive signatures*/
int countDb = minHashDb.Count;
int countQuery = minHashQuery.Count;
int minHashSignatureLen = minHashDb[0].Length;
int similarMinHashValues = 0;
for (int i = 0; i < countDb; i++)
{
for (int j = 0; j < countQuery; j++)
{
for (int k = 0; k < minHashSignatureLen; k++)
if (minHashDb[i][k] == minHashQuery[j][k])
similarMinHashValues++;
}
}
results.Results.SumIdenticalMinHash = similarMinHashValues;
results.Results.AverageIdenticalMinHash = (double) similarMinHashValues/(countDb*countQuery*minHashSignatureLen);
/*Group min hash signatures into LSH Buckets*/
List<Dictionary<int, long>> lshBucketsDb =
minHashDb.Select(item => minHash.GroupMinHashToLSHBuckets(item, hashTables, hashKeys)).ToList();
List<Dictionary<int, long>> lshBucketsQuery =
minHashQuery.Select(item => minHash.GroupMinHashToLSHBuckets(item, hashTables, hashKeys)).ToList();
int countSignatures = lshBucketsDb.Count;
sum = 0;
foreach (Dictionary<int, long> a in lshBucketsDb)
{
Dictionary<int, long>.ValueCollection aValues = a.Values;
foreach (Dictionary<int, long> b in lshBucketsQuery)
{
Dictionary<int, long>.ValueCollection bValues = b.Values;
hashesCount += aValues.Intersect(bValues).Count();
}
}
results.Results.SumJaqLSHBucketSimilarity = -1;
results.Results.AverageJaqLSHBucketSimilarity = -1;
results.Results.TotalIdenticalLSHBuckets = hashesCount;
}
public void ReadPermutationsTest()
{
IPermutations perms = new DbPermutations(connectionstring);
int[][] actual = perms.GetPermutations();
Assert.IsNull(actual);
}
