private ConcurrentDictionary <IModelReference, ResultEntryAccumulator> GetSimilaritiesUsingNonBatchedStrategy(IEnumerable <HashedFingerprint> queryFingerprints, QueryConfiguration configuration, IModelService modelService)
{
var hammingSimilarities = new ConcurrentDictionary <IModelReference, ResultEntryAccumulator>();
foreach (var queryFingerprint in queryFingerprints)
{
var subFingerprints = modelService.ReadSubFingerprints(queryFingerprint.HashBins, configuration);
similarityUtility.AccumulateHammingSimilarity(subFingerprints, queryFingerprint, hammingSimilarities);
}
return(hammingSimilarities);
}
private ConcurrentDictionary <IModelReference, ResultEntryAccumulator> GetSimilaritiesUsingNonBatchedStrategy(IEnumerable <HashedFingerprint> queryFingerprints, QueryConfiguration configuration, IModelService modelService)
{
var hammingSimilarities = new ConcurrentDictionary <IModelReference, ResultEntryAccumulator>();
Parallel.ForEach(queryFingerprints, queryFingerprint =>
{
var subFingerprints = modelService.ReadSubFingerprints(queryFingerprint.HashBins, configuration);
similarityUtility.AccumulateHammingSimilarity(subFingerprints, queryFingerprint, hammingSimilarities, configuration.FingerprintConfiguration.HashingConfig.NumberOfMinHashesPerTable);
});
return(hammingSimilarities);
}
public void ShouldSumUpHammingDistanceAccrossTracks()
{
var hammingSimilarities = new ConcurrentDictionary <IModelReference, ResultEntryAccumulator>();
long[] hashes1 = GenericHashBuckets();
hashes1[0] = 0;
long[] hashes2 = GenericHashBuckets();
long[] hashes3 = GenericHashBuckets();
similarityUtility.AccumulateHammingSimilarity(
new List <SubFingerprintData>
{
new SubFingerprintData(
hashes1,
1,
0d,
new ModelReference <int>(1),
new ModelReference <int>(1)),
new SubFingerprintData(
hashes2,
2,
1.48d,
new ModelReference <int>(1),
new ModelReference <int>(2)),
new SubFingerprintData(
hashes3,
3,
2.92d,
new ModelReference <int>(2),
new ModelReference <int>(2))
},
new HashedFingerprint(GenericSignature(), GenericHashBuckets(), 0, 0, Enumerable.Empty <string>()),
hammingSimilarities);
Assert.AreEqual(2, hammingSimilarities.Count);
Assert.AreEqual(49, hammingSimilarities[new ModelReference <int>(1)].HammingSimilaritySum);
Assert.AreEqual(100, hammingSimilarities[new ModelReference <int>(2)].HammingSimilaritySum);
}
public void ShouldCalculateSimilaritiesCorrectly()
{
var hammingSimilarities = new ConcurrentDictionary <IModelReference, ResultEntryAccumulator>();
var totalCount = 10000;
var countPerTrack = 10;
var tracksCount = totalCount / countPerTrack;
var subFingerprints = GetSubFingerprints(totalCount, countPerTrack);
var queryFingerprint = new HashedFingerprint(new int[25], 0, 0, Enumerable.Empty <string>());
Parallel.ForEach(subFingerprints, subFingerprint => {
var toCompare = new List <SubFingerprintData>(new[] { subFingerprint });
similarityUtility.AccumulateHammingSimilarity(toCompare, queryFingerprint, hammingSimilarities, 4);
});
Assert.AreEqual(tracksCount, hammingSimilarities.Count);
foreach (var pair in hammingSimilarities)
{
Assert.AreEqual(countPerTrack, pair.Value.Matches.Count);
Assert.AreEqual(countPerTrack * 100, pair.Value.HammingSimilaritySum);
Assert.AreEqual(100, pair.Value.BestMatch.HammingSimilarity);
}
}
