/// <summary>
/// Add an item to the estimator.
/// </summary>
/// <param name="item">The entity to add</param>
/// <remarks>based upon the strata, the value is either added to an IBF or to the b-bit minwise estimator.</remarks>
public void Add(TEntity item)
{
var idHash = _configuration.IdHash(_configuration.GetId(item));
var entityHash = _configuration.EntityHash(item);
if (!_strataEstimator.ConditionalAdd(idHash, entityHash))
{
if (_minwiseEstimator == null)
{
_minwiseReplacementCount++;
}
else
{
_minwiseEstimator.Add(new KeyValuePair <int, int>(idHash, entityHash));
}
}
}
/// <summary>
/// Constructor
/// </summary>
/// <param name="configuration">The configuration</param>
/// <param name="bitSize">The number of bits to store per hash</param>
/// <param name="hashCount">The number of hash functions to use.</param>
/// <param name="capacity">The capacity (should be a close approximation of the number of elements added)</param>
/// <remarks>By using bitSize = 1 or bitSize = 2, the accuracy is decreased, thus the hashCount needs to be increased. However, when resemblance is not too small, for example > 0.5, bitSize = 1 can yield similar results as bitSize = 64 with only 3 times the hash count.</remarks>
public BitMinwiseHashEstimator(
IInvertibleBloomFilterConfiguration <TEntity, TId, int, TCount> configuration,
byte bitSize,
int hashCount,
long capacity)
{
_hashCount = hashCount;
_configuration = configuration;
_hashFunctions = GenerateHashes();
_bitSize = bitSize;
_capacity = _configuration.FoldingStrategy?.ComputeFoldableSize(capacity, 0) ?? capacity;
_entityHash = e => unchecked ((int)(ulong)(_configuration.EntityHash(e) + configuration.IdHash(_configuration.GetId(e))));
_slots = new Lazy <int[]>(() => GetMinHashSlots(_hashCount, _capacity));
}
